Commit | Line | Data |
---|---|---|
e8cd7eae GS |
1 | =head1 NAME |
2 | ||
3 | perlhack - How to hack at the Perl internals | |
4 | ||
5 | =head1 DESCRIPTION | |
6 | ||
7 | This document attempts to explain how Perl development takes place, | |
8 | and ends with some suggestions for people wanting to become bona fide | |
9 | porters. | |
10 | ||
11 | The perl5-porters mailing list is where the Perl standard distribution | |
12 | is maintained and developed. The list can get anywhere from 10 to 150 | |
13 | messages a day, depending on the heatedness of the debate. Most days | |
14 | there are two or three patches, extensions, features, or bugs being | |
15 | discussed at a time. | |
16 | ||
f8e3975a | 17 | A searchable archive of the list is at either: |
e8cd7eae GS |
18 | |
19 | http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/ | |
20 | ||
f8e3975a IP |
21 | or |
22 | ||
23 | http://archive.develooper.com/perl5-porters@perl.org/ | |
24 | ||
e8cd7eae GS |
25 | List subscribers (the porters themselves) come in several flavours. |
26 | Some are quiet curious lurkers, who rarely pitch in and instead watch | |
27 | the ongoing development to ensure they're forewarned of new changes or | |
28 | features in Perl. Some are representatives of vendors, who are there | |
29 | to make sure that Perl continues to compile and work on their | |
30 | platforms. Some patch any reported bug that they know how to fix, | |
31 | some are actively patching their pet area (threads, Win32, the regexp | |
32 | engine), while others seem to do nothing but complain. In other | |
33 | words, it's your usual mix of technical people. | |
34 | ||
35 | Over this group of porters presides Larry Wall. He has the final word | |
f6c51b38 | 36 | in what does and does not change in the Perl language. Various |
b432a672 AL |
37 | releases of Perl are shepherded by a "pumpking", a porter |
38 | responsible for gathering patches, deciding on a patch-by-patch, | |
f6c51b38 | 39 | feature-by-feature basis what will and will not go into the release. |
caf100c0 | 40 | For instance, Gurusamy Sarathy was the pumpking for the 5.6 release of |
961f29c6 | 41 | Perl, and Jarkko Hietaniemi was the pumpking for the 5.8 release, and |
1a88dbf8 | 42 | Rafael Garcia-Suarez holds the pumpking crown for the 5.10 release. |
e8cd7eae GS |
43 | |
44 | In addition, various people are pumpkings for different things. For | |
961f29c6 MB |
45 | instance, Andy Dougherty and Jarkko Hietaniemi did a grand job as the |
46 | I<Configure> pumpkin up till the 5.8 release. For the 5.10 release | |
47 | H.Merijn Brand took over. | |
e8cd7eae GS |
48 | |
49 | Larry sees Perl development along the lines of the US government: | |
50 | there's the Legislature (the porters), the Executive branch (the | |
51 | pumpkings), and the Supreme Court (Larry). The legislature can | |
52 | discuss and submit patches to the executive branch all they like, but | |
53 | the executive branch is free to veto them. Rarely, the Supreme Court | |
54 | will side with the executive branch over the legislature, or the | |
55 | legislature over the executive branch. Mostly, however, the | |
56 | legislature and the executive branch are supposed to get along and | |
57 | work out their differences without impeachment or court cases. | |
58 | ||
59 | You might sometimes see reference to Rule 1 and Rule 2. Larry's power | |
60 | as Supreme Court is expressed in The Rules: | |
61 | ||
62 | =over 4 | |
63 | ||
64 | =item 1 | |
65 | ||
66 | Larry is always by definition right about how Perl should behave. | |
67 | This means he has final veto power on the core functionality. | |
68 | ||
69 | =item 2 | |
70 | ||
71 | Larry is allowed to change his mind about any matter at a later date, | |
72 | regardless of whether he previously invoked Rule 1. | |
73 | ||
74 | =back | |
75 | ||
76 | Got that? Larry is always right, even when he was wrong. It's rare | |
77 | to see either Rule exercised, but they are often alluded to. | |
78 | ||
79 | New features and extensions to the language are contentious, because | |
80 | the criteria used by the pumpkings, Larry, and other porters to decide | |
81 | which features should be implemented and incorporated are not codified | |
82 | in a few small design goals as with some other languages. Instead, | |
83 | the heuristics are flexible and often difficult to fathom. Here is | |
84 | one person's list, roughly in decreasing order of importance, of | |
85 | heuristics that new features have to be weighed against: | |
86 | ||
87 | =over 4 | |
88 | ||
89 | =item Does concept match the general goals of Perl? | |
90 | ||
91 | These haven't been written anywhere in stone, but one approximation | |
92 | is: | |
93 | ||
94 | 1. Keep it fast, simple, and useful. | |
95 | 2. Keep features/concepts as orthogonal as possible. | |
96 | 3. No arbitrary limits (platforms, data sizes, cultures). | |
97 | 4. Keep it open and exciting to use/patch/advocate Perl everywhere. | |
98 | 5. Either assimilate new technologies, or build bridges to them. | |
99 | ||
100 | =item Where is the implementation? | |
101 | ||
102 | All the talk in the world is useless without an implementation. In | |
103 | almost every case, the person or people who argue for a new feature | |
104 | will be expected to be the ones who implement it. Porters capable | |
105 | of coding new features have their own agendas, and are not available | |
106 | to implement your (possibly good) idea. | |
107 | ||
108 | =item Backwards compatibility | |
109 | ||
110 | It's a cardinal sin to break existing Perl programs. New warnings are | |
111 | contentious--some say that a program that emits warnings is not | |
112 | broken, while others say it is. Adding keywords has the potential to | |
113 | break programs, changing the meaning of existing token sequences or | |
114 | functions might break programs. | |
115 | ||
116 | =item Could it be a module instead? | |
117 | ||
118 | Perl 5 has extension mechanisms, modules and XS, specifically to avoid | |
119 | the need to keep changing the Perl interpreter. You can write modules | |
120 | that export functions, you can give those functions prototypes so they | |
121 | can be called like built-in functions, you can even write XS code to | |
122 | mess with the runtime data structures of the Perl interpreter if you | |
123 | want to implement really complicated things. If it can be done in a | |
124 | module instead of in the core, it's highly unlikely to be added. | |
125 | ||
126 | =item Is the feature generic enough? | |
127 | ||
128 | Is this something that only the submitter wants added to the language, | |
129 | or would it be broadly useful? Sometimes, instead of adding a feature | |
130 | with a tight focus, the porters might decide to wait until someone | |
131 | implements the more generalized feature. For instance, instead of | |
b432a672 | 132 | implementing a "delayed evaluation" feature, the porters are waiting |
e8cd7eae GS |
133 | for a macro system that would permit delayed evaluation and much more. |
134 | ||
135 | =item Does it potentially introduce new bugs? | |
136 | ||
137 | Radical rewrites of large chunks of the Perl interpreter have the | |
138 | potential to introduce new bugs. The smaller and more localized the | |
139 | change, the better. | |
140 | ||
141 | =item Does it preclude other desirable features? | |
142 | ||
143 | A patch is likely to be rejected if it closes off future avenues of | |
144 | development. For instance, a patch that placed a true and final | |
145 | interpretation on prototypes is likely to be rejected because there | |
146 | are still options for the future of prototypes that haven't been | |
147 | addressed. | |
148 | ||
149 | =item Is the implementation robust? | |
150 | ||
151 | Good patches (tight code, complete, correct) stand more chance of | |
152 | going in. Sloppy or incorrect patches might be placed on the back | |
153 | burner until the pumpking has time to fix, or might be discarded | |
154 | altogether without further notice. | |
155 | ||
156 | =item Is the implementation generic enough to be portable? | |
157 | ||
158 | The worst patches make use of a system-specific features. It's highly | |
353c6505 | 159 | unlikely that non-portable additions to the Perl language will be |
e8cd7eae GS |
160 | accepted. |
161 | ||
a936dd3c NC |
162 | =item Is the implementation tested? |
163 | ||
164 | Patches which change behaviour (fixing bugs or introducing new features) | |
165 | must include regression tests to verify that everything works as expected. | |
166 | Without tests provided by the original author, how can anyone else changing | |
167 | perl in the future be sure that they haven't unwittingly broken the behaviour | |
168 | the patch implements? And without tests, how can the patch's author be | |
9d077eaa | 169 | confident that his/her hard work put into the patch won't be accidentally |
a936dd3c NC |
170 | thrown away by someone in the future? |
171 | ||
e8cd7eae GS |
172 | =item Is there enough documentation? |
173 | ||
174 | Patches without documentation are probably ill-thought out or | |
175 | incomplete. Nothing can be added without documentation, so submitting | |
176 | a patch for the appropriate manpages as well as the source code is | |
a936dd3c | 177 | always a good idea. |
e8cd7eae GS |
178 | |
179 | =item Is there another way to do it? | |
180 | ||
b432a672 AL |
181 | Larry said "Although the Perl Slogan is I<There's More Than One Way |
182 | to Do It>, I hesitate to make 10 ways to do something". This is a | |
e8cd7eae GS |
183 | tricky heuristic to navigate, though--one man's essential addition is |
184 | another man's pointless cruft. | |
185 | ||
186 | =item Does it create too much work? | |
187 | ||
188 | Work for the pumpking, work for Perl programmers, work for module | |
189 | authors, ... Perl is supposed to be easy. | |
190 | ||
f6c51b38 GS |
191 | =item Patches speak louder than words |
192 | ||
193 | Working code is always preferred to pie-in-the-sky ideas. A patch to | |
194 | add a feature stands a much higher chance of making it to the language | |
195 | than does a random feature request, no matter how fervently argued the | |
b432a672 | 196 | request might be. This ties into "Will it be useful?", as the fact |
f6c51b38 GS |
197 | that someone took the time to make the patch demonstrates a strong |
198 | desire for the feature. | |
199 | ||
e8cd7eae GS |
200 | =back |
201 | ||
b432a672 AL |
202 | If you're on the list, you might hear the word "core" bandied |
203 | around. It refers to the standard distribution. "Hacking on the | |
204 | core" means you're changing the C source code to the Perl | |
205 | interpreter. "A core module" is one that ships with Perl. | |
e8cd7eae | 206 | |
a1f349fd MB |
207 | =head2 Keeping in sync |
208 | ||
e8cd7eae | 209 | The source code to the Perl interpreter, in its different versions, is |
b16c2e4a RGS |
210 | kept in a repository managed by the git revision control system. The |
211 | pumpkings and a few others have write access to the repository to check in | |
212 | changes. | |
2be4c08b | 213 | |
b16c2e4a | 214 | How to clone and use the git perl repository is described in L<perlrepository>. |
2be4c08b | 215 | |
b16c2e4a | 216 | You can also choose to use rsync to get a copy of the current source tree |
fe749c9a | 217 | for the bleadperl branch and all maintenance branches : |
0cfb3454 | 218 | |
b16c2e4a RGS |
219 | $ rsync -avz rsync://perl5.git.perl.org/APC/perl-current . |
220 | $ rsync -avz rsync://perl5.git.perl.org/APC/perl-5.10.x . | |
221 | $ rsync -avz rsync://perl5.git.perl.org/APC/perl-5.8.x . | |
222 | $ rsync -avz rsync://perl5.git.perl.org/APC/perl-5.6.x . | |
223 | $ rsync -avz rsync://perl5.git.perl.org/APC/perl-5.005xx . | |
224 | ||
225 | (Add the C<--delete> option to remove leftover files) | |
0cfb3454 GS |
226 | |
227 | You may also want to subscribe to the perl5-changes mailing list to | |
228 | receive a copy of each patch that gets submitted to the maintenance | |
229 | and development "branches" of the perl repository. See | |
230 | http://lists.perl.org/ for subscription information. | |
231 | ||
a1f349fd MB |
232 | If you are a member of the perl5-porters mailing list, it is a good |
233 | thing to keep in touch with the most recent changes. If not only to | |
234 | verify if what you would have posted as a bug report isn't already | |
235 | solved in the most recent available perl development branch, also | |
236 | known as perl-current, bleading edge perl, bleedperl or bleadperl. | |
2be4c08b GS |
237 | |
238 | Needless to say, the source code in perl-current is usually in a perpetual | |
239 | state of evolution. You should expect it to be very buggy. Do B<not> use | |
240 | it for any purpose other than testing and development. | |
e8cd7eae | 241 | |
3fd28c4e | 242 | =head2 Perlbug administration |
52315700 | 243 | |
902821cc RGS |
244 | There is a single remote administrative interface for modifying bug status, |
245 | category, open issues etc. using the B<RT> bugtracker system, maintained | |
246 | by Robert Spier. Become an administrator, and close any bugs you can get | |
3fd28c4e | 247 | your sticky mitts on: |
52315700 | 248 | |
39417508 | 249 | http://bugs.perl.org/ |
52315700 | 250 | |
3fd28c4e | 251 | To email the bug system administrators: |
52315700 | 252 | |
3fd28c4e | 253 | "perlbug-admin" <perlbug-admin@perl.org> |
52315700 | 254 | |
a1f349fd MB |
255 | =head2 Submitting patches |
256 | ||
f7e1e956 MS |
257 | Always submit patches to I<perl5-porters@perl.org>. If you're |
258 | patching a core module and there's an author listed, send the author a | |
259 | copy (see L<Patching a core module>). This lets other porters review | |
260 | your patch, which catches a surprising number of errors in patches. | |
b16c2e4a RGS |
261 | Please patch against the latest B<development> version. (e.g., even if |
262 | you're fixing a bug in the 5.8 track, patch against the C<blead> branch in | |
263 | the git repository.) | |
824d470b RGS |
264 | |
265 | If changes are accepted, they are applied to the development branch. Then | |
fe749c9a | 266 | the maintenance pumpking decides which of those patches is to be |
b16c2e4a RGS |
267 | backported to the maint branch. Only patches that survive the heat of the |
268 | development branch get applied to maintenance versions. | |
f7e1e956 | 269 | |
2c8694a7 JH |
270 | Your patch should update the documentation and test suite. See |
271 | L<Writing a test>. If you have added or removed files in the distribution, | |
272 | edit the MANIFEST file accordingly, sort the MANIFEST file using | |
273 | C<make manisort>, and include those changes as part of your patch. | |
e8cd7eae | 274 | |
824d470b RGS |
275 | Patching documentation also follows the same order: if accepted, a patch |
276 | is first applied to B<development>, and if relevant then it's backported | |
277 | to B<maintenance>. (With an exception for some patches that document | |
278 | behaviour that only appears in the maintenance branch, but which has | |
279 | changed in the development version.) | |
280 | ||
e8cd7eae GS |
281 | To report a bug in Perl, use the program I<perlbug> which comes with |
282 | Perl (if you can't get Perl to work, send mail to the address | |
f18956b7 | 283 | I<perlbug@perl.org> or I<perlbug@perl.com>). Reporting bugs through |
e8cd7eae | 284 | I<perlbug> feeds into the automated bug-tracking system, access to |
902821cc | 285 | which is provided through the web at http://rt.perl.org/rt3/ . It |
e8cd7eae GS |
286 | often pays to check the archives of the perl5-porters mailing list to |
287 | see whether the bug you're reporting has been reported before, and if | |
288 | so whether it was considered a bug. See above for the location of | |
289 | the searchable archives. | |
290 | ||
f224927c | 291 | The CPAN testers ( http://testers.cpan.org/ ) are a group of |
ba139f7d | 292 | volunteers who test CPAN modules on a variety of platforms. Perl |
d3e8af89 GS |
293 | Smokers ( http://www.nntp.perl.org/group/perl.daily-build and |
294 | http://www.nntp.perl.org/group/perl.daily-build.reports/ ) | |
902821cc | 295 | automatically test Perl source releases on platforms with various |
d3e8af89 GS |
296 | configurations. Both efforts welcome volunteers. In order to get |
297 | involved in smoke testing of the perl itself visit | |
298 | L<http://search.cpan.org/dist/Test-Smoke>. In order to start smoke | |
68cbce50 CBW |
299 | testing CPAN modules visit L<http://search.cpan.org/dist/CPANPLUS-YACSmoke/> |
300 | or L<http://search.cpan.org/dist/minismokebox/> or | |
d3e8af89 | 301 | L<http://search.cpan.org/dist/CPAN-Reporter/>. |
e8cd7eae | 302 | |
e8cd7eae GS |
303 | It's a good idea to read and lurk for a while before chipping in. |
304 | That way you'll get to see the dynamic of the conversations, learn the | |
305 | personalities of the players, and hopefully be better prepared to make | |
306 | a useful contribution when do you speak up. | |
307 | ||
308 | If after all this you still think you want to join the perl5-porters | |
f6c51b38 GS |
309 | mailing list, send mail to I<perl5-porters-subscribe@perl.org>. To |
310 | unsubscribe, send mail to I<perl5-porters-unsubscribe@perl.org>. | |
e8cd7eae | 311 | |
a422fd2d SC |
312 | To hack on the Perl guts, you'll need to read the following things: |
313 | ||
314 | =over 3 | |
315 | ||
316 | =item L<perlguts> | |
317 | ||
318 | This is of paramount importance, since it's the documentation of what | |
319 | goes where in the Perl source. Read it over a couple of times and it | |
320 | might start to make sense - don't worry if it doesn't yet, because the | |
321 | best way to study it is to read it in conjunction with poking at Perl | |
322 | source, and we'll do that later on. | |
323 | ||
0aa6d4a5 RU |
324 | Gisle Aas's "illustrated perlguts", also known as I<illguts>, has very |
325 | helpful pictures: | |
de10be12 | 326 | |
0aa6d4a5 | 327 | L<http://search.cpan.org/dist/illguts/> |
a422fd2d SC |
328 | |
329 | =item L<perlxstut> and L<perlxs> | |
330 | ||
331 | A working knowledge of XSUB programming is incredibly useful for core | |
332 | hacking; XSUBs use techniques drawn from the PP code, the portion of the | |
333 | guts that actually executes a Perl program. It's a lot gentler to learn | |
334 | those techniques from simple examples and explanation than from the core | |
335 | itself. | |
336 | ||
337 | =item L<perlapi> | |
338 | ||
339 | The documentation for the Perl API explains what some of the internal | |
340 | functions do, as well as the many macros used in the source. | |
341 | ||
342 | =item F<Porting/pumpkin.pod> | |
343 | ||
344 | This is a collection of words of wisdom for a Perl porter; some of it is | |
345 | only useful to the pumpkin holder, but most of it applies to anyone | |
346 | wanting to go about Perl development. | |
347 | ||
348 | =item The perl5-porters FAQ | |
349 | ||
902821cc RGS |
350 | This should be available from http://dev.perl.org/perl5/docs/p5p-faq.html . |
351 | It contains hints on reading perl5-porters, information on how | |
352 | perl5-porters works and how Perl development in general works. | |
a422fd2d SC |
353 | |
354 | =back | |
355 | ||
356 | =head2 Finding Your Way Around | |
357 | ||
358 | Perl maintenance can be split into a number of areas, and certain people | |
359 | (pumpkins) will have responsibility for each area. These areas sometimes | |
360 | correspond to files or directories in the source kit. Among the areas are: | |
361 | ||
362 | =over 3 | |
363 | ||
364 | =item Core modules | |
365 | ||
c53fdc5e RF |
366 | Modules shipped as part of the Perl core live in various subdirectories, where |
367 | two are dedicated to core-only modules, and two are for the dual-life modules | |
368 | which live on CPAN and may be maintained separately with respect to the Perl | |
369 | core: | |
370 | ||
371 | lib/ is for pure-Perl modules, which exist in the core only. | |
372 | ||
373 | ext/ is for XS extensions, and modules with special Makefile.PL requirements, which exist in the core only. | |
374 | ||
375 | cpan/ is for dual-life modules, where the CPAN module is canonical (should be patched first). | |
376 | ||
377 | dist/ is for dual-life modules, where the blead source is canonical. | |
a422fd2d | 378 | |
f7e1e956 MS |
379 | =item Tests |
380 | ||
381 | There are tests for nearly all the modules, built-ins and major bits | |
382 | of functionality. Test files all have a .t suffix. Module tests live | |
383 | in the F<lib/> and F<ext/> directories next to the module being | |
384 | tested. Others live in F<t/>. See L<Writing a test> | |
385 | ||
a422fd2d SC |
386 | =item Documentation |
387 | ||
388 | Documentation maintenance includes looking after everything in the | |
389 | F<pod/> directory, (as well as contributing new documentation) and | |
390 | the documentation to the modules in core. | |
391 | ||
392 | =item Configure | |
393 | ||
99c47ece | 394 | The Configure process is the way we make Perl portable across the |
a422fd2d | 395 | myriad of operating systems it supports. Responsibility for the |
99c47ece MB |
396 | Configure, build and installation process, as well as the overall |
397 | portability of the core code rests with the Configure pumpkin - | |
398 | others help out with individual operating systems. | |
399 | ||
e1020413 | 400 | The three files that fall under his/her responsibility are Configure, |
99c47ece MB |
401 | config_h.SH, and Porting/Glossary (and a whole bunch of small related |
402 | files that are less important here). The Configure pumpkin decides how | |
403 | patches to these are dealt with. Currently, the Configure pumpkin will | |
404 | accept patches in most common formats, even directly to these files. | |
405 | Other committers are allowed to commit to these files under the strict | |
406 | condition that they will inform the Configure pumpkin, either on IRC | |
407 | (if he/she happens to be around) or through (personal) e-mail. | |
a422fd2d SC |
408 | |
409 | The files involved are the operating system directories, (F<win32/>, | |
410 | F<os2/>, F<vms/> and so on) the shell scripts which generate F<config.h> | |
411 | and F<Makefile>, as well as the metaconfig files which generate | |
412 | F<Configure>. (metaconfig isn't included in the core distribution.) | |
413 | ||
99c47ece MB |
414 | See http://perl5.git.perl.org/metaconfig.git/blob/HEAD:/README for a |
415 | description of the full process involved. | |
416 | ||
a422fd2d SC |
417 | =item Interpreter |
418 | ||
419 | And of course, there's the core of the Perl interpreter itself. Let's | |
420 | have a look at that in a little more detail. | |
421 | ||
422 | =back | |
423 | ||
424 | Before we leave looking at the layout, though, don't forget that | |
425 | F<MANIFEST> contains not only the file names in the Perl distribution, | |
426 | but short descriptions of what's in them, too. For an overview of the | |
427 | important files, try this: | |
428 | ||
429 | perl -lne 'print if /^[^\/]+\.[ch]\s+/' MANIFEST | |
430 | ||
431 | =head2 Elements of the interpreter | |
432 | ||
433 | The work of the interpreter has two main stages: compiling the code | |
434 | into the internal representation, or bytecode, and then executing it. | |
435 | L<perlguts/Compiled code> explains exactly how the compilation stage | |
436 | happens. | |
437 | ||
438 | Here is a short breakdown of perl's operation: | |
439 | ||
440 | =over 3 | |
441 | ||
442 | =item Startup | |
443 | ||
444 | The action begins in F<perlmain.c>. (or F<miniperlmain.c> for miniperl) | |
445 | This is very high-level code, enough to fit on a single screen, and it | |
446 | resembles the code found in L<perlembed>; most of the real action takes | |
447 | place in F<perl.c> | |
448 | ||
9df8f87f LB |
449 | F<perlmain.c> is generated by L<writemain> from F<miniperlmain.c> at |
450 | make time, so you should make perl to follow this along. | |
451 | ||
a422fd2d | 452 | First, F<perlmain.c> allocates some memory and constructs a Perl |
9df8f87f | 453 | interpreter, along these lines: |
a422fd2d SC |
454 | |
455 | 1 PERL_SYS_INIT3(&argc,&argv,&env); | |
456 | 2 | |
457 | 3 if (!PL_do_undump) { | |
458 | 4 my_perl = perl_alloc(); | |
459 | 5 if (!my_perl) | |
460 | 6 exit(1); | |
461 | 7 perl_construct(my_perl); | |
462 | 8 PL_perl_destruct_level = 0; | |
463 | 9 } | |
464 | ||
465 | Line 1 is a macro, and its definition is dependent on your operating | |
466 | system. Line 3 references C<PL_do_undump>, a global variable - all | |
467 | global variables in Perl start with C<PL_>. This tells you whether the | |
468 | current running program was created with the C<-u> flag to perl and then | |
469 | F<undump>, which means it's going to be false in any sane context. | |
470 | ||
471 | Line 4 calls a function in F<perl.c> to allocate memory for a Perl | |
472 | interpreter. It's quite a simple function, and the guts of it looks like | |
473 | this: | |
474 | ||
475 | my_perl = (PerlInterpreter*)PerlMem_malloc(sizeof(PerlInterpreter)); | |
476 | ||
477 | Here you see an example of Perl's system abstraction, which we'll see | |
478 | later: C<PerlMem_malloc> is either your system's C<malloc>, or Perl's | |
479 | own C<malloc> as defined in F<malloc.c> if you selected that option at | |
480 | configure time. | |
481 | ||
9df8f87f LB |
482 | Next, in line 7, we construct the interpreter using perl_construct, |
483 | also in F<perl.c>; this sets up all the special variables that Perl | |
484 | needs, the stacks, and so on. | |
a422fd2d SC |
485 | |
486 | Now we pass Perl the command line options, and tell it to go: | |
487 | ||
488 | exitstatus = perl_parse(my_perl, xs_init, argc, argv, (char **)NULL); | |
9df8f87f LB |
489 | if (!exitstatus) |
490 | perl_run(my_perl); | |
491 | ||
492 | exitstatus = perl_destruct(my_perl); | |
a422fd2d | 493 | |
9df8f87f | 494 | perl_free(my_perl); |
a422fd2d SC |
495 | |
496 | C<perl_parse> is actually a wrapper around C<S_parse_body>, as defined | |
497 | in F<perl.c>, which processes the command line options, sets up any | |
498 | statically linked XS modules, opens the program and calls C<yyparse> to | |
499 | parse it. | |
500 | ||
501 | =item Parsing | |
502 | ||
503 | The aim of this stage is to take the Perl source, and turn it into an op | |
504 | tree. We'll see what one of those looks like later. Strictly speaking, | |
505 | there's three things going on here. | |
506 | ||
507 | C<yyparse>, the parser, lives in F<perly.c>, although you're better off | |
508 | reading the original YACC input in F<perly.y>. (Yes, Virginia, there | |
509 | B<is> a YACC grammar for Perl!) The job of the parser is to take your | |
b432a672 | 510 | code and "understand" it, splitting it into sentences, deciding which |
a422fd2d SC |
511 | operands go with which operators and so on. |
512 | ||
513 | The parser is nobly assisted by the lexer, which chunks up your input | |
514 | into tokens, and decides what type of thing each token is: a variable | |
515 | name, an operator, a bareword, a subroutine, a core function, and so on. | |
516 | The main point of entry to the lexer is C<yylex>, and that and its | |
517 | associated routines can be found in F<toke.c>. Perl isn't much like | |
518 | other computer languages; it's highly context sensitive at times, it can | |
519 | be tricky to work out what sort of token something is, or where a token | |
520 | ends. As such, there's a lot of interplay between the tokeniser and the | |
521 | parser, which can get pretty frightening if you're not used to it. | |
522 | ||
523 | As the parser understands a Perl program, it builds up a tree of | |
524 | operations for the interpreter to perform during execution. The routines | |
525 | which construct and link together the various operations are to be found | |
526 | in F<op.c>, and will be examined later. | |
527 | ||
528 | =item Optimization | |
529 | ||
530 | Now the parsing stage is complete, and the finished tree represents | |
531 | the operations that the Perl interpreter needs to perform to execute our | |
532 | program. Next, Perl does a dry run over the tree looking for | |
533 | optimisations: constant expressions such as C<3 + 4> will be computed | |
534 | now, and the optimizer will also see if any multiple operations can be | |
535 | replaced with a single one. For instance, to fetch the variable C<$foo>, | |
536 | instead of grabbing the glob C<*foo> and looking at the scalar | |
537 | component, the optimizer fiddles the op tree to use a function which | |
538 | directly looks up the scalar in question. The main optimizer is C<peep> | |
539 | in F<op.c>, and many ops have their own optimizing functions. | |
540 | ||
541 | =item Running | |
542 | ||
543 | Now we're finally ready to go: we have compiled Perl byte code, and all | |
544 | that's left to do is run it. The actual execution is done by the | |
545 | C<runops_standard> function in F<run.c>; more specifically, it's done by | |
546 | these three innocent looking lines: | |
547 | ||
548 | while ((PL_op = CALL_FPTR(PL_op->op_ppaddr)(aTHX))) { | |
549 | PERL_ASYNC_CHECK(); | |
550 | } | |
551 | ||
552 | You may be more comfortable with the Perl version of that: | |
553 | ||
554 | PERL_ASYNC_CHECK() while $Perl::op = &{$Perl::op->{function}}; | |
555 | ||
556 | Well, maybe not. Anyway, each op contains a function pointer, which | |
557 | stipulates the function which will actually carry out the operation. | |
558 | This function will return the next op in the sequence - this allows for | |
559 | things like C<if> which choose the next op dynamically at run time. | |
560 | The C<PERL_ASYNC_CHECK> makes sure that things like signals interrupt | |
561 | execution if required. | |
562 | ||
563 | The actual functions called are known as PP code, and they're spread | |
b432a672 | 564 | between four files: F<pp_hot.c> contains the "hot" code, which is most |
a422fd2d SC |
565 | often used and highly optimized, F<pp_sys.c> contains all the |
566 | system-specific functions, F<pp_ctl.c> contains the functions which | |
567 | implement control structures (C<if>, C<while> and the like) and F<pp.c> | |
568 | contains everything else. These are, if you like, the C code for Perl's | |
569 | built-in functions and operators. | |
570 | ||
dfc98234 DM |
571 | Note that each C<pp_> function is expected to return a pointer to the next |
572 | op. Calls to perl subs (and eval blocks) are handled within the same | |
573 | runops loop, and do not consume extra space on the C stack. For example, | |
574 | C<pp_entersub> and C<pp_entertry> just push a C<CxSUB> or C<CxEVAL> block | |
575 | struct onto the context stack which contain the address of the op | |
576 | following the sub call or eval. They then return the first op of that sub | |
577 | or eval block, and so execution continues of that sub or block. Later, a | |
578 | C<pp_leavesub> or C<pp_leavetry> op pops the C<CxSUB> or C<CxEVAL>, | |
579 | retrieves the return op from it, and returns it. | |
580 | ||
581 | =item Exception handing | |
582 | ||
0503309d | 583 | Perl's exception handing (i.e. C<die> etc.) is built on top of the low-level |
dfc98234 | 584 | C<setjmp()>/C<longjmp()> C-library functions. These basically provide a |
28a5cf3b | 585 | way to capture the current PC and SP registers and later restore them; i.e. |
dfc98234 DM |
586 | a C<longjmp()> continues at the point in code where a previous C<setjmp()> |
587 | was done, with anything further up on the C stack being lost. This is why | |
588 | code should always save values using C<SAVE_FOO> rather than in auto | |
589 | variables. | |
590 | ||
591 | The perl core wraps C<setjmp()> etc in the macros C<JMPENV_PUSH> and | |
592 | C<JMPENV_JUMP>. The basic rule of perl exceptions is that C<exit>, and | |
593 | C<die> (in the absence of C<eval>) perform a C<JMPENV_JUMP(2)>, while | |
594 | C<die> within C<eval> does a C<JMPENV_JUMP(3)>. | |
595 | ||
596 | At entry points to perl, such as C<perl_parse()>, C<perl_run()> and | |
597 | C<call_sv(cv, G_EVAL)> each does a C<JMPENV_PUSH>, then enter a runops | |
598 | loop or whatever, and handle possible exception returns. For a 2 return, | |
599 | final cleanup is performed, such as popping stacks and calling C<CHECK> or | |
600 | C<END> blocks. Amongst other things, this is how scope cleanup still | |
601 | occurs during an C<exit>. | |
602 | ||
603 | If a C<die> can find a C<CxEVAL> block on the context stack, then the | |
604 | stack is popped to that level and the return op in that block is assigned | |
605 | to C<PL_restartop>; then a C<JMPENV_JUMP(3)> is performed. This normally | |
606 | passes control back to the guard. In the case of C<perl_run> and | |
607 | C<call_sv>, a non-null C<PL_restartop> triggers re-entry to the runops | |
608 | loop. The is the normal way that C<die> or C<croak> is handled within an | |
609 | C<eval>. | |
610 | ||
611 | Sometimes ops are executed within an inner runops loop, such as tie, sort | |
612 | or overload code. In this case, something like | |
613 | ||
614 | sub FETCH { eval { die } } | |
615 | ||
616 | would cause a longjmp right back to the guard in C<perl_run>, popping both | |
617 | runops loops, which is clearly incorrect. One way to avoid this is for the | |
618 | tie code to do a C<JMPENV_PUSH> before executing C<FETCH> in the inner | |
619 | runops loop, but for efficiency reasons, perl in fact just sets a flag, | |
620 | using C<CATCH_SET(TRUE)>. The C<pp_require>, C<pp_entereval> and | |
621 | C<pp_entertry> ops check this flag, and if true, they call C<docatch>, | |
622 | which does a C<JMPENV_PUSH> and starts a new runops level to execute the | |
623 | code, rather than doing it on the current loop. | |
624 | ||
625 | As a further optimisation, on exit from the eval block in the C<FETCH>, | |
626 | execution of the code following the block is still carried on in the inner | |
627 | loop. When an exception is raised, C<docatch> compares the C<JMPENV> | |
628 | level of the C<CxEVAL> with C<PL_top_env> and if they differ, just | |
629 | re-throws the exception. In this way any inner loops get popped. | |
630 | ||
631 | Here's an example. | |
632 | ||
633 | 1: eval { tie @a, 'A' }; | |
634 | 2: sub A::TIEARRAY { | |
635 | 3: eval { die }; | |
636 | 4: die; | |
637 | 5: } | |
638 | ||
639 | To run this code, C<perl_run> is called, which does a C<JMPENV_PUSH> then | |
640 | enters a runops loop. This loop executes the eval and tie ops on line 1, | |
641 | with the eval pushing a C<CxEVAL> onto the context stack. | |
642 | ||
643 | The C<pp_tie> does a C<CATCH_SET(TRUE)>, then starts a second runops loop | |
644 | to execute the body of C<TIEARRAY>. When it executes the entertry op on | |
645 | line 3, C<CATCH_GET> is true, so C<pp_entertry> calls C<docatch> which | |
646 | does a C<JMPENV_PUSH> and starts a third runops loop, which then executes | |
647 | the die op. At this point the C call stack looks like this: | |
648 | ||
649 | Perl_pp_die | |
650 | Perl_runops # third loop | |
651 | S_docatch_body | |
652 | S_docatch | |
653 | Perl_pp_entertry | |
654 | Perl_runops # second loop | |
655 | S_call_body | |
656 | Perl_call_sv | |
657 | Perl_pp_tie | |
658 | Perl_runops # first loop | |
659 | S_run_body | |
660 | perl_run | |
661 | main | |
662 | ||
663 | and the context and data stacks, as shown by C<-Dstv>, look like: | |
664 | ||
665 | STACK 0: MAIN | |
666 | CX 0: BLOCK => | |
667 | CX 1: EVAL => AV() PV("A"\0) | |
668 | retop=leave | |
669 | STACK 1: MAGIC | |
670 | CX 0: SUB => | |
671 | retop=(null) | |
672 | CX 1: EVAL => * | |
673 | retop=nextstate | |
674 | ||
675 | The die pops the first C<CxEVAL> off the context stack, sets | |
676 | C<PL_restartop> from it, does a C<JMPENV_JUMP(3)>, and control returns to | |
677 | the top C<docatch>. This then starts another third-level runops level, | |
678 | which executes the nextstate, pushmark and die ops on line 4. At the point | |
679 | that the second C<pp_die> is called, the C call stack looks exactly like | |
680 | that above, even though we are no longer within an inner eval; this is | |
681 | because of the optimization mentioned earlier. However, the context stack | |
682 | now looks like this, ie with the top CxEVAL popped: | |
683 | ||
684 | STACK 0: MAIN | |
685 | CX 0: BLOCK => | |
686 | CX 1: EVAL => AV() PV("A"\0) | |
687 | retop=leave | |
688 | STACK 1: MAGIC | |
689 | CX 0: SUB => | |
690 | retop=(null) | |
691 | ||
692 | The die on line 4 pops the context stack back down to the CxEVAL, leaving | |
693 | it as: | |
694 | ||
695 | STACK 0: MAIN | |
696 | CX 0: BLOCK => | |
697 | ||
698 | As usual, C<PL_restartop> is extracted from the C<CxEVAL>, and a | |
699 | C<JMPENV_JUMP(3)> done, which pops the C stack back to the docatch: | |
700 | ||
701 | S_docatch | |
702 | Perl_pp_entertry | |
703 | Perl_runops # second loop | |
704 | S_call_body | |
705 | Perl_call_sv | |
706 | Perl_pp_tie | |
707 | Perl_runops # first loop | |
708 | S_run_body | |
709 | perl_run | |
710 | main | |
711 | ||
712 | In this case, because the C<JMPENV> level recorded in the C<CxEVAL> | |
713 | differs from the current one, C<docatch> just does a C<JMPENV_JUMP(3)> | |
714 | and the C stack unwinds to: | |
715 | ||
716 | perl_run | |
717 | main | |
718 | ||
719 | Because C<PL_restartop> is non-null, C<run_body> starts a new runops loop | |
720 | and execution continues. | |
721 | ||
a422fd2d SC |
722 | =back |
723 | ||
724 | =head2 Internal Variable Types | |
725 | ||
726 | You should by now have had a look at L<perlguts>, which tells you about | |
727 | Perl's internal variable types: SVs, HVs, AVs and the rest. If not, do | |
728 | that now. | |
729 | ||
730 | These variables are used not only to represent Perl-space variables, but | |
731 | also any constants in the code, as well as some structures completely | |
732 | internal to Perl. The symbol table, for instance, is an ordinary Perl | |
733 | hash. Your code is represented by an SV as it's read into the parser; | |
734 | any program files you call are opened via ordinary Perl filehandles, and | |
735 | so on. | |
736 | ||
737 | The core L<Devel::Peek|Devel::Peek> module lets us examine SVs from a | |
738 | Perl program. Let's see, for instance, how Perl treats the constant | |
739 | C<"hello">. | |
740 | ||
741 | % perl -MDevel::Peek -e 'Dump("hello")' | |
742 | 1 SV = PV(0xa041450) at 0xa04ecbc | |
743 | 2 REFCNT = 1 | |
744 | 3 FLAGS = (POK,READONLY,pPOK) | |
745 | 4 PV = 0xa0484e0 "hello"\0 | |
746 | 5 CUR = 5 | |
747 | 6 LEN = 6 | |
748 | ||
749 | Reading C<Devel::Peek> output takes a bit of practise, so let's go | |
750 | through it line by line. | |
751 | ||
752 | Line 1 tells us we're looking at an SV which lives at C<0xa04ecbc> in | |
753 | memory. SVs themselves are very simple structures, but they contain a | |
754 | pointer to a more complex structure. In this case, it's a PV, a | |
755 | structure which holds a string value, at location C<0xa041450>. Line 2 | |
756 | is the reference count; there are no other references to this data, so | |
757 | it's 1. | |
758 | ||
759 | Line 3 are the flags for this SV - it's OK to use it as a PV, it's a | |
760 | read-only SV (because it's a constant) and the data is a PV internally. | |
761 | Next we've got the contents of the string, starting at location | |
762 | C<0xa0484e0>. | |
763 | ||
764 | Line 5 gives us the current length of the string - note that this does | |
765 | B<not> include the null terminator. Line 6 is not the length of the | |
766 | string, but the length of the currently allocated buffer; as the string | |
767 | grows, Perl automatically extends the available storage via a routine | |
768 | called C<SvGROW>. | |
769 | ||
770 | You can get at any of these quantities from C very easily; just add | |
771 | C<Sv> to the name of the field shown in the snippet, and you've got a | |
772 | macro which will return the value: C<SvCUR(sv)> returns the current | |
773 | length of the string, C<SvREFCOUNT(sv)> returns the reference count, | |
774 | C<SvPV(sv, len)> returns the string itself with its length, and so on. | |
775 | More macros to manipulate these properties can be found in L<perlguts>. | |
776 | ||
777 | Let's take an example of manipulating a PV, from C<sv_catpvn>, in F<sv.c> | |
778 | ||
779 | 1 void | |
780 | 2 Perl_sv_catpvn(pTHX_ register SV *sv, register const char *ptr, register STRLEN len) | |
781 | 3 { | |
782 | 4 STRLEN tlen; | |
783 | 5 char *junk; | |
784 | ||
785 | 6 junk = SvPV_force(sv, tlen); | |
786 | 7 SvGROW(sv, tlen + len + 1); | |
787 | 8 if (ptr == junk) | |
788 | 9 ptr = SvPVX(sv); | |
789 | 10 Move(ptr,SvPVX(sv)+tlen,len,char); | |
790 | 11 SvCUR(sv) += len; | |
791 | 12 *SvEND(sv) = '\0'; | |
792 | 13 (void)SvPOK_only_UTF8(sv); /* validate pointer */ | |
793 | 14 SvTAINT(sv); | |
794 | 15 } | |
795 | ||
796 | This is a function which adds a string, C<ptr>, of length C<len> onto | |
797 | the end of the PV stored in C<sv>. The first thing we do in line 6 is | |
798 | make sure that the SV B<has> a valid PV, by calling the C<SvPV_force> | |
799 | macro to force a PV. As a side effect, C<tlen> gets set to the current | |
800 | value of the PV, and the PV itself is returned to C<junk>. | |
801 | ||
b1866b2d | 802 | In line 7, we make sure that the SV will have enough room to accommodate |
a422fd2d SC |
803 | the old string, the new string and the null terminator. If C<LEN> isn't |
804 | big enough, C<SvGROW> will reallocate space for us. | |
805 | ||
806 | Now, if C<junk> is the same as the string we're trying to add, we can | |
807 | grab the string directly from the SV; C<SvPVX> is the address of the PV | |
808 | in the SV. | |
809 | ||
810 | Line 10 does the actual catenation: the C<Move> macro moves a chunk of | |
811 | memory around: we move the string C<ptr> to the end of the PV - that's | |
812 | the start of the PV plus its current length. We're moving C<len> bytes | |
813 | of type C<char>. After doing so, we need to tell Perl we've extended the | |
814 | string, by altering C<CUR> to reflect the new length. C<SvEND> is a | |
815 | macro which gives us the end of the string, so that needs to be a | |
816 | C<"\0">. | |
817 | ||
818 | Line 13 manipulates the flags; since we've changed the PV, any IV or NV | |
819 | values will no longer be valid: if we have C<$a=10; $a.="6";> we don't | |
1e54db1a | 820 | want to use the old IV of 10. C<SvPOK_only_utf8> is a special UTF-8-aware |
a422fd2d SC |
821 | version of C<SvPOK_only>, a macro which turns off the IOK and NOK flags |
822 | and turns on POK. The final C<SvTAINT> is a macro which launders tainted | |
823 | data if taint mode is turned on. | |
824 | ||
825 | AVs and HVs are more complicated, but SVs are by far the most common | |
826 | variable type being thrown around. Having seen something of how we | |
827 | manipulate these, let's go on and look at how the op tree is | |
828 | constructed. | |
829 | ||
830 | =head2 Op Trees | |
831 | ||
832 | First, what is the op tree, anyway? The op tree is the parsed | |
833 | representation of your program, as we saw in our section on parsing, and | |
834 | it's the sequence of operations that Perl goes through to execute your | |
835 | program, as we saw in L</Running>. | |
836 | ||
837 | An op is a fundamental operation that Perl can perform: all the built-in | |
838 | functions and operators are ops, and there are a series of ops which | |
839 | deal with concepts the interpreter needs internally - entering and | |
840 | leaving a block, ending a statement, fetching a variable, and so on. | |
841 | ||
842 | The op tree is connected in two ways: you can imagine that there are two | |
843 | "routes" through it, two orders in which you can traverse the tree. | |
844 | First, parse order reflects how the parser understood the code, and | |
845 | secondly, execution order tells perl what order to perform the | |
846 | operations in. | |
847 | ||
848 | The easiest way to examine the op tree is to stop Perl after it has | |
849 | finished parsing, and get it to dump out the tree. This is exactly what | |
7d7d5695 RGS |
850 | the compiler backends L<B::Terse|B::Terse>, L<B::Concise|B::Concise> |
851 | and L<B::Debug|B::Debug> do. | |
a422fd2d SC |
852 | |
853 | Let's have a look at how Perl sees C<$a = $b + $c>: | |
854 | ||
855 | % perl -MO=Terse -e '$a=$b+$c' | |
856 | 1 LISTOP (0x8179888) leave | |
857 | 2 OP (0x81798b0) enter | |
858 | 3 COP (0x8179850) nextstate | |
859 | 4 BINOP (0x8179828) sassign | |
860 | 5 BINOP (0x8179800) add [1] | |
861 | 6 UNOP (0x81796e0) null [15] | |
862 | 7 SVOP (0x80fafe0) gvsv GV (0x80fa4cc) *b | |
863 | 8 UNOP (0x81797e0) null [15] | |
864 | 9 SVOP (0x8179700) gvsv GV (0x80efeb0) *c | |
865 | 10 UNOP (0x816b4f0) null [15] | |
866 | 11 SVOP (0x816dcf0) gvsv GV (0x80fa460) *a | |
867 | ||
868 | Let's start in the middle, at line 4. This is a BINOP, a binary | |
869 | operator, which is at location C<0x8179828>. The specific operator in | |
870 | question is C<sassign> - scalar assignment - and you can find the code | |
871 | which implements it in the function C<pp_sassign> in F<pp_hot.c>. As a | |
872 | binary operator, it has two children: the add operator, providing the | |
873 | result of C<$b+$c>, is uppermost on line 5, and the left hand side is on | |
874 | line 10. | |
875 | ||
876 | Line 10 is the null op: this does exactly nothing. What is that doing | |
877 | there? If you see the null op, it's a sign that something has been | |
878 | optimized away after parsing. As we mentioned in L</Optimization>, | |
879 | the optimization stage sometimes converts two operations into one, for | |
880 | example when fetching a scalar variable. When this happens, instead of | |
881 | rewriting the op tree and cleaning up the dangling pointers, it's easier | |
882 | just to replace the redundant operation with the null op. Originally, | |
883 | the tree would have looked like this: | |
884 | ||
885 | 10 SVOP (0x816b4f0) rv2sv [15] | |
886 | 11 SVOP (0x816dcf0) gv GV (0x80fa460) *a | |
887 | ||
888 | That is, fetch the C<a> entry from the main symbol table, and then look | |
889 | at the scalar component of it: C<gvsv> (C<pp_gvsv> into F<pp_hot.c>) | |
890 | happens to do both these things. | |
891 | ||
892 | The right hand side, starting at line 5 is similar to what we've just | |
893 | seen: we have the C<add> op (C<pp_add> also in F<pp_hot.c>) add together | |
894 | two C<gvsv>s. | |
895 | ||
896 | Now, what's this about? | |
897 | ||
898 | 1 LISTOP (0x8179888) leave | |
899 | 2 OP (0x81798b0) enter | |
900 | 3 COP (0x8179850) nextstate | |
901 | ||
902 | C<enter> and C<leave> are scoping ops, and their job is to perform any | |
903 | housekeeping every time you enter and leave a block: lexical variables | |
904 | are tidied up, unreferenced variables are destroyed, and so on. Every | |
905 | program will have those first three lines: C<leave> is a list, and its | |
906 | children are all the statements in the block. Statements are delimited | |
907 | by C<nextstate>, so a block is a collection of C<nextstate> ops, with | |
908 | the ops to be performed for each statement being the children of | |
909 | C<nextstate>. C<enter> is a single op which functions as a marker. | |
910 | ||
911 | That's how Perl parsed the program, from top to bottom: | |
912 | ||
913 | Program | |
914 | | | |
915 | Statement | |
916 | | | |
917 | = | |
918 | / \ | |
919 | / \ | |
920 | $a + | |
921 | / \ | |
922 | $b $c | |
923 | ||
924 | However, it's impossible to B<perform> the operations in this order: | |
925 | you have to find the values of C<$b> and C<$c> before you add them | |
926 | together, for instance. So, the other thread that runs through the op | |
927 | tree is the execution order: each op has a field C<op_next> which points | |
928 | to the next op to be run, so following these pointers tells us how perl | |
929 | executes the code. We can traverse the tree in this order using | |
930 | the C<exec> option to C<B::Terse>: | |
931 | ||
932 | % perl -MO=Terse,exec -e '$a=$b+$c' | |
933 | 1 OP (0x8179928) enter | |
934 | 2 COP (0x81798c8) nextstate | |
935 | 3 SVOP (0x81796c8) gvsv GV (0x80fa4d4) *b | |
936 | 4 SVOP (0x8179798) gvsv GV (0x80efeb0) *c | |
937 | 5 BINOP (0x8179878) add [1] | |
938 | 6 SVOP (0x816dd38) gvsv GV (0x80fa468) *a | |
939 | 7 BINOP (0x81798a0) sassign | |
940 | 8 LISTOP (0x8179900) leave | |
941 | ||
942 | This probably makes more sense for a human: enter a block, start a | |
943 | statement. Get the values of C<$b> and C<$c>, and add them together. | |
944 | Find C<$a>, and assign one to the other. Then leave. | |
945 | ||
946 | The way Perl builds up these op trees in the parsing process can be | |
947 | unravelled by examining F<perly.y>, the YACC grammar. Let's take the | |
948 | piece we need to construct the tree for C<$a = $b + $c> | |
949 | ||
950 | 1 term : term ASSIGNOP term | |
951 | 2 { $$ = newASSIGNOP(OPf_STACKED, $1, $2, $3); } | |
952 | 3 | term ADDOP term | |
953 | 4 { $$ = newBINOP($2, 0, scalar($1), scalar($3)); } | |
954 | ||
955 | If you're not used to reading BNF grammars, this is how it works: You're | |
956 | fed certain things by the tokeniser, which generally end up in upper | |
957 | case. Here, C<ADDOP>, is provided when the tokeniser sees C<+> in your | |
958 | code. C<ASSIGNOP> is provided when C<=> is used for assigning. These are | |
b432a672 | 959 | "terminal symbols", because you can't get any simpler than them. |
a422fd2d SC |
960 | |
961 | The grammar, lines one and three of the snippet above, tells you how to | |
b432a672 | 962 | build up more complex forms. These complex forms, "non-terminal symbols" |
a422fd2d SC |
963 | are generally placed in lower case. C<term> here is a non-terminal |
964 | symbol, representing a single expression. | |
965 | ||
966 | The grammar gives you the following rule: you can make the thing on the | |
967 | left of the colon if you see all the things on the right in sequence. | |
968 | This is called a "reduction", and the aim of parsing is to completely | |
969 | reduce the input. There are several different ways you can perform a | |
970 | reduction, separated by vertical bars: so, C<term> followed by C<=> | |
971 | followed by C<term> makes a C<term>, and C<term> followed by C<+> | |
972 | followed by C<term> can also make a C<term>. | |
973 | ||
974 | So, if you see two terms with an C<=> or C<+>, between them, you can | |
975 | turn them into a single expression. When you do this, you execute the | |
976 | code in the block on the next line: if you see C<=>, you'll do the code | |
977 | in line 2. If you see C<+>, you'll do the code in line 4. It's this code | |
978 | which contributes to the op tree. | |
979 | ||
980 | | term ADDOP term | |
981 | { $$ = newBINOP($2, 0, scalar($1), scalar($3)); } | |
982 | ||
983 | What this does is creates a new binary op, and feeds it a number of | |
984 | variables. The variables refer to the tokens: C<$1> is the first token in | |
985 | the input, C<$2> the second, and so on - think regular expression | |
986 | backreferences. C<$$> is the op returned from this reduction. So, we | |
987 | call C<newBINOP> to create a new binary operator. The first parameter to | |
988 | C<newBINOP>, a function in F<op.c>, is the op type. It's an addition | |
989 | operator, so we want the type to be C<ADDOP>. We could specify this | |
990 | directly, but it's right there as the second token in the input, so we | |
b432a672 AL |
991 | use C<$2>. The second parameter is the op's flags: 0 means "nothing |
992 | special". Then the things to add: the left and right hand side of our | |
a422fd2d SC |
993 | expression, in scalar context. |
994 | ||
995 | =head2 Stacks | |
996 | ||
997 | When perl executes something like C<addop>, how does it pass on its | |
998 | results to the next op? The answer is, through the use of stacks. Perl | |
999 | has a number of stacks to store things it's currently working on, and | |
1000 | we'll look at the three most important ones here. | |
1001 | ||
1002 | =over 3 | |
1003 | ||
1004 | =item Argument stack | |
1005 | ||
1006 | Arguments are passed to PP code and returned from PP code using the | |
1007 | argument stack, C<ST>. The typical way to handle arguments is to pop | |
1008 | them off the stack, deal with them how you wish, and then push the result | |
1009 | back onto the stack. This is how, for instance, the cosine operator | |
1010 | works: | |
1011 | ||
1012 | NV value; | |
1013 | value = POPn; | |
1014 | value = Perl_cos(value); | |
1015 | XPUSHn(value); | |
1016 | ||
1017 | We'll see a more tricky example of this when we consider Perl's macros | |
1018 | below. C<POPn> gives you the NV (floating point value) of the top SV on | |
1019 | the stack: the C<$x> in C<cos($x)>. Then we compute the cosine, and push | |
1020 | the result back as an NV. The C<X> in C<XPUSHn> means that the stack | |
1021 | should be extended if necessary - it can't be necessary here, because we | |
1022 | know there's room for one more item on the stack, since we've just | |
1023 | removed one! The C<XPUSH*> macros at least guarantee safety. | |
1024 | ||
1025 | Alternatively, you can fiddle with the stack directly: C<SP> gives you | |
1026 | the first element in your portion of the stack, and C<TOP*> gives you | |
1027 | the top SV/IV/NV/etc. on the stack. So, for instance, to do unary | |
1028 | negation of an integer: | |
1029 | ||
1030 | SETi(-TOPi); | |
1031 | ||
1032 | Just set the integer value of the top stack entry to its negation. | |
1033 | ||
1034 | Argument stack manipulation in the core is exactly the same as it is in | |
1035 | XSUBs - see L<perlxstut>, L<perlxs> and L<perlguts> for a longer | |
1036 | description of the macros used in stack manipulation. | |
1037 | ||
1038 | =item Mark stack | |
1039 | ||
b432a672 | 1040 | I say "your portion of the stack" above because PP code doesn't |
a422fd2d SC |
1041 | necessarily get the whole stack to itself: if your function calls |
1042 | another function, you'll only want to expose the arguments aimed for the | |
1043 | called function, and not (necessarily) let it get at your own data. The | |
b432a672 | 1044 | way we do this is to have a "virtual" bottom-of-stack, exposed to each |
a422fd2d SC |
1045 | function. The mark stack keeps bookmarks to locations in the argument |
1046 | stack usable by each function. For instance, when dealing with a tied | |
b432a672 | 1047 | variable, (internally, something with "P" magic) Perl has to call |
a422fd2d SC |
1048 | methods for accesses to the tied variables. However, we need to separate |
1049 | the arguments exposed to the method to the argument exposed to the | |
ed233832 DM |
1050 | original function - the store or fetch or whatever it may be. Here's |
1051 | roughly how the tied C<push> is implemented; see C<av_push> in F<av.c>: | |
a422fd2d SC |
1052 | |
1053 | 1 PUSHMARK(SP); | |
1054 | 2 EXTEND(SP,2); | |
1055 | 3 PUSHs(SvTIED_obj((SV*)av, mg)); | |
1056 | 4 PUSHs(val); | |
1057 | 5 PUTBACK; | |
1058 | 6 ENTER; | |
1059 | 7 call_method("PUSH", G_SCALAR|G_DISCARD); | |
1060 | 8 LEAVE; | |
13a2d996 | 1061 | |
a422fd2d SC |
1062 | Let's examine the whole implementation, for practice: |
1063 | ||
1064 | 1 PUSHMARK(SP); | |
1065 | ||
1066 | Push the current state of the stack pointer onto the mark stack. This is | |
1067 | so that when we've finished adding items to the argument stack, Perl | |
1068 | knows how many things we've added recently. | |
1069 | ||
1070 | 2 EXTEND(SP,2); | |
1071 | 3 PUSHs(SvTIED_obj((SV*)av, mg)); | |
1072 | 4 PUSHs(val); | |
1073 | ||
1074 | We're going to add two more items onto the argument stack: when you have | |
1075 | a tied array, the C<PUSH> subroutine receives the object and the value | |
1076 | to be pushed, and that's exactly what we have here - the tied object, | |
1077 | retrieved with C<SvTIED_obj>, and the value, the SV C<val>. | |
1078 | ||
1079 | 5 PUTBACK; | |
1080 | ||
e89a6d4e JD |
1081 | Next we tell Perl to update the global stack pointer from our internal |
1082 | variable: C<dSP> only gave us a local copy, not a reference to the global. | |
a422fd2d SC |
1083 | |
1084 | 6 ENTER; | |
1085 | 7 call_method("PUSH", G_SCALAR|G_DISCARD); | |
1086 | 8 LEAVE; | |
1087 | ||
1088 | C<ENTER> and C<LEAVE> localise a block of code - they make sure that all | |
1089 | variables are tidied up, everything that has been localised gets | |
1090 | its previous value returned, and so on. Think of them as the C<{> and | |
1091 | C<}> of a Perl block. | |
1092 | ||
1093 | To actually do the magic method call, we have to call a subroutine in | |
1094 | Perl space: C<call_method> takes care of that, and it's described in | |
1095 | L<perlcall>. We call the C<PUSH> method in scalar context, and we're | |
e89a6d4e JD |
1096 | going to discard its return value. The call_method() function |
1097 | removes the top element of the mark stack, so there is nothing for | |
1098 | the caller to clean up. | |
a422fd2d | 1099 | |
a422fd2d SC |
1100 | =item Save stack |
1101 | ||
1102 | C doesn't have a concept of local scope, so perl provides one. We've | |
1103 | seen that C<ENTER> and C<LEAVE> are used as scoping braces; the save | |
1104 | stack implements the C equivalent of, for example: | |
1105 | ||
1106 | { | |
1107 | local $foo = 42; | |
1108 | ... | |
1109 | } | |
1110 | ||
1111 | See L<perlguts/Localising Changes> for how to use the save stack. | |
1112 | ||
1113 | =back | |
1114 | ||
1115 | =head2 Millions of Macros | |
1116 | ||
1117 | One thing you'll notice about the Perl source is that it's full of | |
1118 | macros. Some have called the pervasive use of macros the hardest thing | |
1119 | to understand, others find it adds to clarity. Let's take an example, | |
1120 | the code which implements the addition operator: | |
1121 | ||
1122 | 1 PP(pp_add) | |
1123 | 2 { | |
39644a26 | 1124 | 3 dSP; dATARGET; tryAMAGICbin(add,opASSIGN); |
a422fd2d SC |
1125 | 4 { |
1126 | 5 dPOPTOPnnrl_ul; | |
1127 | 6 SETn( left + right ); | |
1128 | 7 RETURN; | |
1129 | 8 } | |
1130 | 9 } | |
1131 | ||
1132 | Every line here (apart from the braces, of course) contains a macro. The | |
1133 | first line sets up the function declaration as Perl expects for PP code; | |
1134 | line 3 sets up variable declarations for the argument stack and the | |
1135 | target, the return value of the operation. Finally, it tries to see if | |
1136 | the addition operation is overloaded; if so, the appropriate subroutine | |
1137 | is called. | |
1138 | ||
1139 | Line 5 is another variable declaration - all variable declarations start | |
1140 | with C<d> - which pops from the top of the argument stack two NVs (hence | |
1141 | C<nn>) and puts them into the variables C<right> and C<left>, hence the | |
1142 | C<rl>. These are the two operands to the addition operator. Next, we | |
1143 | call C<SETn> to set the NV of the return value to the result of adding | |
1144 | the two values. This done, we return - the C<RETURN> macro makes sure | |
1145 | that our return value is properly handled, and we pass the next operator | |
1146 | to run back to the main run loop. | |
1147 | ||
1148 | Most of these macros are explained in L<perlapi>, and some of the more | |
1149 | important ones are explained in L<perlxs> as well. Pay special attention | |
1150 | to L<perlguts/Background and PERL_IMPLICIT_CONTEXT> for information on | |
1151 | the C<[pad]THX_?> macros. | |
1152 | ||
52d59bef JH |
1153 | =head2 The .i Targets |
1154 | ||
1155 | You can expand the macros in a F<foo.c> file by saying | |
1156 | ||
1157 | make foo.i | |
1158 | ||
1159 | which will expand the macros using cpp. Don't be scared by the results. | |
1160 | ||
955fec6b JH |
1161 | =head1 SOURCE CODE STATIC ANALYSIS |
1162 | ||
1163 | Various tools exist for analysing C source code B<statically>, as | |
1164 | opposed to B<dynamically>, that is, without executing the code. | |
1165 | It is possible to detect resource leaks, undefined behaviour, type | |
1166 | mismatches, portability problems, code paths that would cause illegal | |
1167 | memory accesses, and other similar problems by just parsing the C code | |
1168 | and looking at the resulting graph, what does it tell about the | |
1169 | execution and data flows. As a matter of fact, this is exactly | |
1170 | how C compilers know to give warnings about dubious code. | |
1171 | ||
1172 | =head2 lint, splint | |
1173 | ||
1174 | The good old C code quality inspector, C<lint>, is available in | |
1175 | several platforms, but please be aware that there are several | |
1176 | different implementations of it by different vendors, which means that | |
1177 | the flags are not identical across different platforms. | |
1178 | ||
1179 | There is a lint variant called C<splint> (Secure Programming Lint) | |
1180 | available from http://www.splint.org/ that should compile on any | |
1181 | Unix-like platform. | |
1182 | ||
1183 | There are C<lint> and <splint> targets in Makefile, but you may have | |
1184 | to diddle with the flags (see above). | |
1185 | ||
1186 | =head2 Coverity | |
1187 | ||
1188 | Coverity (http://www.coverity.com/) is a product similar to lint and | |
1189 | as a testbed for their product they periodically check several open | |
1190 | source projects, and they give out accounts to open source developers | |
1191 | to the defect databases. | |
1192 | ||
1193 | =head2 cpd (cut-and-paste detector) | |
1194 | ||
1195 | The cpd tool detects cut-and-paste coding. If one instance of the | |
1196 | cut-and-pasted code changes, all the other spots should probably be | |
1197 | changed, too. Therefore such code should probably be turned into a | |
1198 | subroutine or a macro. | |
1199 | ||
1200 | cpd (http://pmd.sourceforge.net/cpd.html) is part of the pmd project | |
1201 | (http://pmd.sourceforge.net/). pmd was originally written for static | |
1202 | analysis of Java code, but later the cpd part of it was extended to | |
1203 | parse also C and C++. | |
1204 | ||
a52aaefa RGS |
1205 | Download the pmd-bin-X.Y.zip () from the SourceForge site, extract the |
1206 | pmd-X.Y.jar from it, and then run that on source code thusly: | |
955fec6b JH |
1207 | |
1208 | java -cp pmd-X.Y.jar net.sourceforge.pmd.cpd.CPD --minimum-tokens 100 --files /some/where/src --language c > cpd.txt | |
1209 | ||
1210 | You may run into memory limits, in which case you should use the -Xmx option: | |
1211 | ||
1212 | java -Xmx512M ... | |
1213 | ||
1214 | =head2 gcc warnings | |
1215 | ||
1216 | Though much can be written about the inconsistency and coverage | |
1217 | problems of gcc warnings (like C<-Wall> not meaning "all the | |
1218 | warnings", or some common portability problems not being covered by | |
1219 | C<-Wall>, or C<-ansi> and C<-pedantic> both being a poorly defined | |
1220 | collection of warnings, and so forth), gcc is still a useful tool in | |
1221 | keeping our coding nose clean. | |
1222 | ||
1223 | The C<-Wall> is by default on. | |
1224 | ||
a8e98a71 JH |
1225 | The C<-ansi> (and its sidekick, C<-pedantic>) would be nice to be on |
1226 | always, but unfortunately they are not safe on all platforms, they can | |
1227 | for example cause fatal conflicts with the system headers (Solaris | |
1228 | being a prime example). If Configure C<-Dgccansipedantic> is used, | |
1229 | the C<cflags> frontend selects C<-ansi -pedantic> for the platforms | |
1230 | where they are known to be safe. | |
955fec6b JH |
1231 | |
1232 | Starting from Perl 5.9.4 the following extra flags are added: | |
1233 | ||
1234 | =over 4 | |
1235 | ||
1236 | =item * | |
1237 | ||
1238 | C<-Wendif-labels> | |
1239 | ||
1240 | =item * | |
1241 | ||
1242 | C<-Wextra> | |
1243 | ||
1244 | =item * | |
1245 | ||
1246 | C<-Wdeclaration-after-statement> | |
1247 | ||
1248 | =back | |
1249 | ||
1250 | The following flags would be nice to have but they would first need | |
0503309d | 1251 | their own Augean stablemaster: |
955fec6b JH |
1252 | |
1253 | =over 4 | |
1254 | ||
1255 | =item * | |
1256 | ||
1257 | C<-Wpointer-arith> | |
1258 | ||
1259 | =item * | |
1260 | ||
1261 | C<-Wshadow> | |
1262 | ||
1263 | =item * | |
1264 | ||
1265 | C<-Wstrict-prototypes> | |
1266 | ||
955fec6b JH |
1267 | =back |
1268 | ||
1269 | The C<-Wtraditional> is another example of the annoying tendency of | |
ac036724 | 1270 | gcc to bundle a lot of warnings under one switch (it would be |
1271 | impossible to deploy in practice because it would complain a lot) but | |
955fec6b JH |
1272 | it does contain some warnings that would be beneficial to have available |
1273 | on their own, such as the warning about string constants inside macros | |
1274 | containing the macro arguments: this behaved differently pre-ANSI | |
1275 | than it does in ANSI, and some C compilers are still in transition, | |
1276 | AIX being an example. | |
1277 | ||
1278 | =head2 Warnings of other C compilers | |
1279 | ||
1280 | Other C compilers (yes, there B<are> other C compilers than gcc) often | |
1281 | have their "strict ANSI" or "strict ANSI with some portability extensions" | |
1282 | modes on, like for example the Sun Workshop has its C<-Xa> mode on | |
1283 | (though implicitly), or the DEC (these days, HP...) has its C<-std1> | |
1284 | mode on. | |
1285 | ||
1286 | =head2 DEBUGGING | |
1287 | ||
1288 | You can compile a special debugging version of Perl, which allows you | |
1289 | to use the C<-D> option of Perl to tell more about what Perl is doing. | |
1290 | But sometimes there is no alternative than to dive in with a debugger, | |
1291 | either to see the stack trace of a core dump (very useful in a bug | |
1292 | report), or trying to figure out what went wrong before the core dump | |
1293 | happened, or how did we end up having wrong or unexpected results. | |
1294 | ||
a422fd2d SC |
1295 | =head2 Poking at Perl |
1296 | ||
1297 | To really poke around with Perl, you'll probably want to build Perl for | |
1298 | debugging, like this: | |
1299 | ||
1300 | ./Configure -d -D optimize=-g | |
1301 | make | |
1302 | ||
1303 | C<-g> is a flag to the C compiler to have it produce debugging | |
955fec6b JH |
1304 | information which will allow us to step through a running program, |
1305 | and to see in which C function we are at (without the debugging | |
1306 | information we might see only the numerical addresses of the functions, | |
1307 | which is not very helpful). | |
1308 | ||
a422fd2d SC |
1309 | F<Configure> will also turn on the C<DEBUGGING> compilation symbol which |
1310 | enables all the internal debugging code in Perl. There are a whole bunch | |
1311 | of things you can debug with this: L<perlrun> lists them all, and the | |
1312 | best way to find out about them is to play about with them. The most | |
1313 | useful options are probably | |
1314 | ||
1315 | l Context (loop) stack processing | |
1316 | t Trace execution | |
1317 | o Method and overloading resolution | |
1318 | c String/numeric conversions | |
1319 | ||
1320 | Some of the functionality of the debugging code can be achieved using XS | |
1321 | modules. | |
13a2d996 | 1322 | |
a422fd2d SC |
1323 | -Dr => use re 'debug' |
1324 | -Dx => use O 'Debug' | |
1325 | ||
1326 | =head2 Using a source-level debugger | |
1327 | ||
1328 | If the debugging output of C<-D> doesn't help you, it's time to step | |
1329 | through perl's execution with a source-level debugger. | |
1330 | ||
1331 | =over 3 | |
1332 | ||
1333 | =item * | |
1334 | ||
955fec6b JH |
1335 | We'll use C<gdb> for our examples here; the principles will apply to |
1336 | any debugger (many vendors call their debugger C<dbx>), but check the | |
1337 | manual of the one you're using. | |
a422fd2d SC |
1338 | |
1339 | =back | |
1340 | ||
1341 | To fire up the debugger, type | |
1342 | ||
1343 | gdb ./perl | |
1344 | ||
955fec6b JH |
1345 | Or if you have a core dump: |
1346 | ||
1347 | gdb ./perl core | |
1348 | ||
a422fd2d SC |
1349 | You'll want to do that in your Perl source tree so the debugger can read |
1350 | the source code. You should see the copyright message, followed by the | |
1351 | prompt. | |
1352 | ||
1353 | (gdb) | |
1354 | ||
1355 | C<help> will get you into the documentation, but here are the most | |
1356 | useful commands: | |
1357 | ||
1358 | =over 3 | |
1359 | ||
1360 | =item run [args] | |
1361 | ||
1362 | Run the program with the given arguments. | |
1363 | ||
1364 | =item break function_name | |
1365 | ||
1366 | =item break source.c:xxx | |
1367 | ||
1368 | Tells the debugger that we'll want to pause execution when we reach | |
cea6626f | 1369 | either the named function (but see L<perlguts/Internal Functions>!) or the given |
a422fd2d SC |
1370 | line in the named source file. |
1371 | ||
1372 | =item step | |
1373 | ||
1374 | Steps through the program a line at a time. | |
1375 | ||
1376 | =item next | |
1377 | ||
1378 | Steps through the program a line at a time, without descending into | |
1379 | functions. | |
1380 | ||
1381 | =item continue | |
1382 | ||
1383 | Run until the next breakpoint. | |
1384 | ||
1385 | =item finish | |
1386 | ||
1387 | Run until the end of the current function, then stop again. | |
1388 | ||
13a2d996 | 1389 | =item 'enter' |
a422fd2d SC |
1390 | |
1391 | Just pressing Enter will do the most recent operation again - it's a | |
1392 | blessing when stepping through miles of source code. | |
1393 | ||
1394 | =item print | |
1395 | ||
1396 | Execute the given C code and print its results. B<WARNING>: Perl makes | |
52d59bef JH |
1397 | heavy use of macros, and F<gdb> does not necessarily support macros |
1398 | (see later L</"gdb macro support">). You'll have to substitute them | |
1399 | yourself, or to invoke cpp on the source code files | |
1400 | (see L</"The .i Targets">) | |
1401 | So, for instance, you can't say | |
a422fd2d SC |
1402 | |
1403 | print SvPV_nolen(sv) | |
1404 | ||
1405 | but you have to say | |
1406 | ||
1407 | print Perl_sv_2pv_nolen(sv) | |
1408 | ||
ffc145e8 RK |
1409 | =back |
1410 | ||
a422fd2d SC |
1411 | You may find it helpful to have a "macro dictionary", which you can |
1412 | produce by saying C<cpp -dM perl.c | sort>. Even then, F<cpp> won't | |
07aa3531 | 1413 | recursively apply those macros for you. |
52d59bef JH |
1414 | |
1415 | =head2 gdb macro support | |
a422fd2d | 1416 | |
52d59bef | 1417 | Recent versions of F<gdb> have fairly good macro support, but |
ea031e66 RGS |
1418 | in order to use it you'll need to compile perl with macro definitions |
1419 | included in the debugging information. Using F<gcc> version 3.1, this | |
1420 | means configuring with C<-Doptimize=-g3>. Other compilers might use a | |
1421 | different switch (if they support debugging macros at all). | |
1422 | ||
a422fd2d SC |
1423 | =head2 Dumping Perl Data Structures |
1424 | ||
1425 | One way to get around this macro hell is to use the dumping functions in | |
1426 | F<dump.c>; these work a little like an internal | |
1427 | L<Devel::Peek|Devel::Peek>, but they also cover OPs and other structures | |
1428 | that you can't get at from Perl. Let's take an example. We'll use the | |
07aa3531 | 1429 | C<$a = $b + $c> we used before, but give it a bit of context: |
a422fd2d SC |
1430 | C<$b = "6XXXX"; $c = 2.3;>. Where's a good place to stop and poke around? |
1431 | ||
1432 | What about C<pp_add>, the function we examined earlier to implement the | |
1433 | C<+> operator: | |
1434 | ||
1435 | (gdb) break Perl_pp_add | |
1436 | Breakpoint 1 at 0x46249f: file pp_hot.c, line 309. | |
1437 | ||
cea6626f | 1438 | Notice we use C<Perl_pp_add> and not C<pp_add> - see L<perlguts/Internal Functions>. |
a422fd2d SC |
1439 | With the breakpoint in place, we can run our program: |
1440 | ||
1441 | (gdb) run -e '$b = "6XXXX"; $c = 2.3; $a = $b + $c' | |
1442 | ||
1443 | Lots of junk will go past as gdb reads in the relevant source files and | |
1444 | libraries, and then: | |
1445 | ||
1446 | Breakpoint 1, Perl_pp_add () at pp_hot.c:309 | |
39644a26 | 1447 | 309 dSP; dATARGET; tryAMAGICbin(add,opASSIGN); |
a422fd2d SC |
1448 | (gdb) step |
1449 | 311 dPOPTOPnnrl_ul; | |
1450 | (gdb) | |
1451 | ||
1452 | We looked at this bit of code before, and we said that C<dPOPTOPnnrl_ul> | |
1453 | arranges for two C<NV>s to be placed into C<left> and C<right> - let's | |
1454 | slightly expand it: | |
1455 | ||
1456 | #define dPOPTOPnnrl_ul NV right = POPn; \ | |
1457 | SV *leftsv = TOPs; \ | |
1458 | NV left = USE_LEFT(leftsv) ? SvNV(leftsv) : 0.0 | |
1459 | ||
1460 | C<POPn> takes the SV from the top of the stack and obtains its NV either | |
1461 | directly (if C<SvNOK> is set) or by calling the C<sv_2nv> function. | |
1462 | C<TOPs> takes the next SV from the top of the stack - yes, C<POPn> uses | |
1463 | C<TOPs> - but doesn't remove it. We then use C<SvNV> to get the NV from | |
07aa3531 | 1464 | C<leftsv> in the same way as before - yes, C<POPn> uses C<SvNV>. |
a422fd2d SC |
1465 | |
1466 | Since we don't have an NV for C<$b>, we'll have to use C<sv_2nv> to | |
1467 | convert it. If we step again, we'll find ourselves there: | |
1468 | ||
1469 | Perl_sv_2nv (sv=0xa0675d0) at sv.c:1669 | |
1470 | 1669 if (!sv) | |
1471 | (gdb) | |
1472 | ||
1473 | We can now use C<Perl_sv_dump> to investigate the SV: | |
1474 | ||
1475 | SV = PV(0xa057cc0) at 0xa0675d0 | |
1476 | REFCNT = 1 | |
1477 | FLAGS = (POK,pPOK) | |
1478 | PV = 0xa06a510 "6XXXX"\0 | |
1479 | CUR = 5 | |
1480 | LEN = 6 | |
1481 | $1 = void | |
1482 | ||
1483 | We know we're going to get C<6> from this, so let's finish the | |
1484 | subroutine: | |
1485 | ||
1486 | (gdb) finish | |
1487 | Run till exit from #0 Perl_sv_2nv (sv=0xa0675d0) at sv.c:1671 | |
1488 | 0x462669 in Perl_pp_add () at pp_hot.c:311 | |
1489 | 311 dPOPTOPnnrl_ul; | |
1490 | ||
1491 | We can also dump out this op: the current op is always stored in | |
1492 | C<PL_op>, and we can dump it with C<Perl_op_dump>. This'll give us | |
1493 | similar output to L<B::Debug|B::Debug>. | |
1494 | ||
1495 | { | |
1496 | 13 TYPE = add ===> 14 | |
1497 | TARG = 1 | |
1498 | FLAGS = (SCALAR,KIDS) | |
1499 | { | |
1500 | TYPE = null ===> (12) | |
1501 | (was rv2sv) | |
1502 | FLAGS = (SCALAR,KIDS) | |
1503 | { | |
1504 | 11 TYPE = gvsv ===> 12 | |
1505 | FLAGS = (SCALAR) | |
1506 | GV = main::b | |
1507 | } | |
1508 | } | |
1509 | ||
10f58044 | 1510 | # finish this later # |
a422fd2d SC |
1511 | |
1512 | =head2 Patching | |
1513 | ||
1514 | All right, we've now had a look at how to navigate the Perl sources and | |
1515 | some things you'll need to know when fiddling with them. Let's now get | |
1516 | on and create a simple patch. Here's something Larry suggested: if a | |
07aa3531 | 1517 | C<U> is the first active format during a C<pack>, (for example, |
a422fd2d | 1518 | C<pack "U3C8", @stuff>) then the resulting string should be treated as |
1e54db1a | 1519 | UTF-8 encoded. |
a422fd2d | 1520 | |
168a53cc DR |
1521 | If you are working with a git clone of the Perl repository, you will want to |
1522 | create a branch for your changes. This will make creating a proper patch much | |
1523 | simpler. See the L<perlrepository> for details on how to do this. | |
1524 | ||
a422fd2d SC |
1525 | How do we prepare to fix this up? First we locate the code in question - |
1526 | the C<pack> happens at runtime, so it's going to be in one of the F<pp> | |
1527 | files. Sure enough, C<pp_pack> is in F<pp.c>. Since we're going to be | |
1528 | altering this file, let's copy it to F<pp.c~>. | |
1529 | ||
a6ec74c1 JH |
1530 | [Well, it was in F<pp.c> when this tutorial was written. It has now been |
1531 | split off with C<pp_unpack> to its own file, F<pp_pack.c>] | |
1532 | ||
a422fd2d SC |
1533 | Now let's look over C<pp_pack>: we take a pattern into C<pat>, and then |
1534 | loop over the pattern, taking each format character in turn into | |
1535 | C<datum_type>. Then for each possible format character, we swallow up | |
1536 | the other arguments in the pattern (a field width, an asterisk, and so | |
1537 | on) and convert the next chunk input into the specified format, adding | |
1538 | it onto the output SV C<cat>. | |
1539 | ||
1540 | How do we know if the C<U> is the first format in the C<pat>? Well, if | |
1541 | we have a pointer to the start of C<pat> then, if we see a C<U> we can | |
1542 | test whether we're still at the start of the string. So, here's where | |
1543 | C<pat> is set up: | |
1544 | ||
1545 | STRLEN fromlen; | |
1546 | register char *pat = SvPVx(*++MARK, fromlen); | |
1547 | register char *patend = pat + fromlen; | |
1548 | register I32 len; | |
1549 | I32 datumtype; | |
1550 | SV *fromstr; | |
1551 | ||
1552 | We'll have another string pointer in there: | |
1553 | ||
1554 | STRLEN fromlen; | |
1555 | register char *pat = SvPVx(*++MARK, fromlen); | |
1556 | register char *patend = pat + fromlen; | |
1557 | + char *patcopy; | |
1558 | register I32 len; | |
1559 | I32 datumtype; | |
1560 | SV *fromstr; | |
1561 | ||
1562 | And just before we start the loop, we'll set C<patcopy> to be the start | |
1563 | of C<pat>: | |
1564 | ||
1565 | items = SP - MARK; | |
1566 | MARK++; | |
1567 | sv_setpvn(cat, "", 0); | |
1568 | + patcopy = pat; | |
1569 | while (pat < patend) { | |
1570 | ||
1571 | Now if we see a C<U> which was at the start of the string, we turn on | |
1e54db1a | 1572 | the C<UTF8> flag for the output SV, C<cat>: |
a422fd2d SC |
1573 | |
1574 | + if (datumtype == 'U' && pat==patcopy+1) | |
1575 | + SvUTF8_on(cat); | |
1576 | if (datumtype == '#') { | |
1577 | while (pat < patend && *pat != '\n') | |
1578 | pat++; | |
1579 | ||
1580 | Remember that it has to be C<patcopy+1> because the first character of | |
1581 | the string is the C<U> which has been swallowed into C<datumtype!> | |
1582 | ||
1583 | Oops, we forgot one thing: what if there are spaces at the start of the | |
1584 | pattern? C<pack(" U*", @stuff)> will have C<U> as the first active | |
1585 | character, even though it's not the first thing in the pattern. In this | |
1586 | case, we have to advance C<patcopy> along with C<pat> when we see spaces: | |
1587 | ||
1588 | if (isSPACE(datumtype)) | |
1589 | continue; | |
1590 | ||
1591 | needs to become | |
1592 | ||
1593 | if (isSPACE(datumtype)) { | |
1594 | patcopy++; | |
1595 | continue; | |
1596 | } | |
1597 | ||
1598 | OK. That's the C part done. Now we must do two additional things before | |
1599 | this patch is ready to go: we've changed the behaviour of Perl, and so | |
1600 | we must document that change. We must also provide some more regression | |
1601 | tests to make sure our patch works and doesn't create a bug somewhere | |
1602 | else along the line. | |
1603 | ||
b23b8711 MS |
1604 | The regression tests for each operator live in F<t/op/>, and so we |
1605 | make a copy of F<t/op/pack.t> to F<t/op/pack.t~>. Now we can add our | |
1606 | tests to the end. First, we'll test that the C<U> does indeed create | |
07aa3531 | 1607 | Unicode strings. |
b23b8711 MS |
1608 | |
1609 | t/op/pack.t has a sensible ok() function, but if it didn't we could | |
35c336e6 | 1610 | use the one from t/test.pl. |
b23b8711 | 1611 | |
35c336e6 MS |
1612 | require './test.pl'; |
1613 | plan( tests => 159 ); | |
b23b8711 MS |
1614 | |
1615 | so instead of this: | |
a422fd2d SC |
1616 | |
1617 | print 'not ' unless "1.20.300.4000" eq sprintf "%vd", pack("U*",1,20,300,4000); | |
1618 | print "ok $test\n"; $test++; | |
1619 | ||
35c336e6 MS |
1620 | we can write the more sensible (see L<Test::More> for a full |
1621 | explanation of is() and other testing functions). | |
b23b8711 | 1622 | |
07aa3531 | 1623 | is( "1.20.300.4000", sprintf "%vd", pack("U*",1,20,300,4000), |
38a44b82 | 1624 | "U* produces Unicode" ); |
b23b8711 | 1625 | |
a422fd2d SC |
1626 | Now we'll test that we got that space-at-the-beginning business right: |
1627 | ||
35c336e6 | 1628 | is( "1.20.300.4000", sprintf "%vd", pack(" U*",1,20,300,4000), |
812f5127 | 1629 | " with spaces at the beginning" ); |
a422fd2d SC |
1630 | |
1631 | And finally we'll test that we don't make Unicode strings if C<U> is B<not> | |
1632 | the first active format: | |
1633 | ||
35c336e6 | 1634 | isnt( v1.20.300.4000, sprintf "%vd", pack("C0U*",1,20,300,4000), |
38a44b82 | 1635 | "U* not first isn't Unicode" ); |
a422fd2d | 1636 | |
35c336e6 MS |
1637 | Mustn't forget to change the number of tests which appears at the top, |
1638 | or else the automated tester will get confused. This will either look | |
1639 | like this: | |
a422fd2d | 1640 | |
35c336e6 MS |
1641 | print "1..156\n"; |
1642 | ||
1643 | or this: | |
1644 | ||
1645 | plan( tests => 156 ); | |
a422fd2d SC |
1646 | |
1647 | We now compile up Perl, and run it through the test suite. Our new | |
1648 | tests pass, hooray! | |
1649 | ||
1650 | Finally, the documentation. The job is never done until the paperwork is | |
1651 | over, so let's describe the change we've just made. The relevant place | |
1652 | is F<pod/perlfunc.pod>; again, we make a copy, and then we'll insert | |
1653 | this text in the description of C<pack>: | |
1654 | ||
1655 | =item * | |
1656 | ||
1657 | If the pattern begins with a C<U>, the resulting string will be treated | |
1e54db1a JH |
1658 | as UTF-8-encoded Unicode. You can force UTF-8 encoding on in a string |
1659 | with an initial C<U0>, and the bytes that follow will be interpreted as | |
1660 | Unicode characters. If you don't want this to happen, you can begin your | |
1661 | pattern with C<C0> (or anything else) to force Perl not to UTF-8 encode your | |
a422fd2d SC |
1662 | string, and then follow this with a C<U*> somewhere in your pattern. |
1663 | ||
f7e1e956 MS |
1664 | =head2 Patching a core module |
1665 | ||
1666 | This works just like patching anything else, with an extra | |
1667 | consideration. Many core modules also live on CPAN. If this is so, | |
1668 | patch the CPAN version instead of the core and send the patch off to | |
1669 | the module maintainer (with a copy to p5p). This will help the module | |
1670 | maintainer keep the CPAN version in sync with the core version without | |
1671 | constantly scanning p5p. | |
1672 | ||
db300100 RGS |
1673 | The list of maintainers of core modules is usefully documented in |
1674 | F<Porting/Maintainers.pl>. | |
1675 | ||
acbe17fc JP |
1676 | =head2 Adding a new function to the core |
1677 | ||
1678 | If, as part of a patch to fix a bug, or just because you have an | |
1679 | especially good idea, you decide to add a new function to the core, | |
1680 | discuss your ideas on p5p well before you start work. It may be that | |
1681 | someone else has already attempted to do what you are considering and | |
1682 | can give lots of good advice or even provide you with bits of code | |
1683 | that they already started (but never finished). | |
1684 | ||
1685 | You have to follow all of the advice given above for patching. It is | |
1686 | extremely important to test any addition thoroughly and add new tests | |
1687 | to explore all boundary conditions that your new function is expected | |
1688 | to handle. If your new function is used only by one module (e.g. toke), | |
1689 | then it should probably be named S_your_function (for static); on the | |
210b36aa | 1690 | other hand, if you expect it to accessible from other functions in |
acbe17fc JP |
1691 | Perl, you should name it Perl_your_function. See L<perlguts/Internal Functions> |
1692 | for more details. | |
1693 | ||
1694 | The location of any new code is also an important consideration. Don't | |
1695 | just create a new top level .c file and put your code there; you would | |
1696 | have to make changes to Configure (so the Makefile is created properly), | |
1697 | as well as possibly lots of include files. This is strictly pumpking | |
1698 | business. | |
1699 | ||
1700 | It is better to add your function to one of the existing top level | |
1701 | source code files, but your choice is complicated by the nature of | |
1702 | the Perl distribution. Only the files that are marked as compiled | |
1703 | static are located in the perl executable. Everything else is located | |
1704 | in the shared library (or DLL if you are running under WIN32). So, | |
1705 | for example, if a function was only used by functions located in | |
1706 | toke.c, then your code can go in toke.c. If, however, you want to call | |
1707 | the function from universal.c, then you should put your code in another | |
1708 | location, for example util.c. | |
1709 | ||
1710 | In addition to writing your c-code, you will need to create an | |
1711 | appropriate entry in embed.pl describing your function, then run | |
1712 | 'make regen_headers' to create the entries in the numerous header | |
1713 | files that perl needs to compile correctly. See L<perlguts/Internal Functions> | |
1714 | for information on the various options that you can set in embed.pl. | |
1715 | You will forget to do this a few (or many) times and you will get | |
1716 | warnings during the compilation phase. Make sure that you mention | |
1717 | this when you post your patch to P5P; the pumpking needs to know this. | |
1718 | ||
1719 | When you write your new code, please be conscious of existing code | |
884bad00 | 1720 | conventions used in the perl source files. See L<perlstyle> for |
acbe17fc JP |
1721 | details. Although most of the guidelines discussed seem to focus on |
1722 | Perl code, rather than c, they all apply (except when they don't ;). | |
552d2b87 DR |
1723 | Also see I<perlrepository> for lots of details about both formatting and |
1724 | submitting patches of your changes. | |
acbe17fc JP |
1725 | |
1726 | Lastly, TEST TEST TEST TEST TEST any code before posting to p5p. | |
1727 | Test on as many platforms as you can find. Test as many perl | |
1728 | Configure options as you can (e.g. MULTIPLICITY). If you have | |
1729 | profiling or memory tools, see L<EXTERNAL TOOLS FOR DEBUGGING PERL> | |
210b36aa | 1730 | below for how to use them to further test your code. Remember that |
acbe17fc JP |
1731 | most of the people on P5P are doing this on their own time and |
1732 | don't have the time to debug your code. | |
f7e1e956 MS |
1733 | |
1734 | =head2 Writing a test | |
1735 | ||
1736 | Every module and built-in function has an associated test file (or | |
1737 | should...). If you add or change functionality, you have to write a | |
1738 | test. If you fix a bug, you have to write a test so that bug never | |
1739 | comes back. If you alter the docs, it would be nice to test what the | |
1740 | new documentation says. | |
1741 | ||
1742 | In short, if you submit a patch you probably also have to patch the | |
1743 | tests. | |
1744 | ||
1745 | For modules, the test file is right next to the module itself. | |
1746 | F<lib/strict.t> tests F<lib/strict.pm>. This is a recent innovation, | |
1747 | so there are some snags (and it would be wonderful for you to brush | |
1748 | them out), but it basically works that way. Everything else lives in | |
1749 | F<t/>. | |
1750 | ||
d5f28025 JV |
1751 | If you add a new test directory under F<t/>, it is imperative that you |
1752 | add that directory to F<t/HARNESS> and F<t/TEST>. | |
1753 | ||
f7e1e956 MS |
1754 | =over 3 |
1755 | ||
1756 | =item F<t/base/> | |
1757 | ||
1758 | Testing of the absolute basic functionality of Perl. Things like | |
1759 | C<if>, basic file reads and writes, simple regexes, etc. These are | |
1760 | run first in the test suite and if any of them fail, something is | |
1761 | I<really> broken. | |
1762 | ||
1763 | =item F<t/cmd/> | |
1764 | ||
1765 | These test the basic control structures, C<if/else>, C<while>, | |
35c336e6 | 1766 | subroutines, etc. |
f7e1e956 MS |
1767 | |
1768 | =item F<t/comp/> | |
1769 | ||
1770 | Tests basic issues of how Perl parses and compiles itself. | |
1771 | ||
1772 | =item F<t/io/> | |
1773 | ||
1774 | Tests for built-in IO functions, including command line arguments. | |
1775 | ||
1776 | =item F<t/lib/> | |
1777 | ||
1778 | The old home for the module tests, you shouldn't put anything new in | |
1779 | here. There are still some bits and pieces hanging around in here | |
1780 | that need to be moved. Perhaps you could move them? Thanks! | |
1781 | ||
3c295041 RGS |
1782 | =item F<t/mro/> |
1783 | ||
0503309d | 1784 | Tests for perl's method resolution order implementations |
3c295041 RGS |
1785 | (see L<mro>). |
1786 | ||
f7e1e956 MS |
1787 | =item F<t/op/> |
1788 | ||
1789 | Tests for perl's built in functions that don't fit into any of the | |
1790 | other directories. | |
1791 | ||
a4499558 YO |
1792 | =item F<t/re/> |
1793 | ||
1794 | Tests for regex related functions or behaviour. (These used to live | |
1795 | in t/op). | |
1796 | ||
f7e1e956 MS |
1797 | =item F<t/run/> |
1798 | ||
1799 | Testing features of how perl actually runs, including exit codes and | |
1800 | handling of PERL* environment variables. | |
1801 | ||
244d9cb7 RGS |
1802 | =item F<t/uni/> |
1803 | ||
1804 | Tests for the core support of Unicode. | |
1805 | ||
1806 | =item F<t/win32/> | |
1807 | ||
1808 | Windows-specific tests. | |
1809 | ||
1810 | =item F<t/x2p> | |
1811 | ||
1812 | A test suite for the s2p converter. | |
1813 | ||
f7e1e956 MS |
1814 | =back |
1815 | ||
1816 | The core uses the same testing style as the rest of Perl, a simple | |
1817 | "ok/not ok" run through Test::Harness, but there are a few special | |
1818 | considerations. | |
1819 | ||
35c336e6 MS |
1820 | There are three ways to write a test in the core. Test::More, |
1821 | t/test.pl and ad hoc C<print $test ? "ok 42\n" : "not ok 42\n">. The | |
1822 | decision of which to use depends on what part of the test suite you're | |
1823 | working on. This is a measure to prevent a high-level failure (such | |
1824 | as Config.pm breaking) from causing basic functionality tests to fail. | |
1825 | ||
07aa3531 | 1826 | =over 4 |
35c336e6 MS |
1827 | |
1828 | =item t/base t/comp | |
1829 | ||
1830 | Since we don't know if require works, or even subroutines, use ad hoc | |
1831 | tests for these two. Step carefully to avoid using the feature being | |
1832 | tested. | |
1833 | ||
1834 | =item t/cmd t/run t/io t/op | |
1835 | ||
1836 | Now that basic require() and subroutines are tested, you can use the | |
1837 | t/test.pl library which emulates the important features of Test::More | |
1838 | while using a minimum of core features. | |
1839 | ||
1840 | You can also conditionally use certain libraries like Config, but be | |
1841 | sure to skip the test gracefully if it's not there. | |
1842 | ||
1843 | =item t/lib ext lib | |
1844 | ||
1845 | Now that the core of Perl is tested, Test::More can be used. You can | |
1846 | also use the full suite of core modules in the tests. | |
1847 | ||
1848 | =back | |
f7e1e956 MS |
1849 | |
1850 | When you say "make test" Perl uses the F<t/TEST> program to run the | |
07aa3531 JC |
1851 | test suite (except under Win32 where it uses F<t/harness> instead.) |
1852 | All tests are run from the F<t/> directory, B<not> the directory | |
1853 | which contains the test. This causes some problems with the tests | |
7205a85d | 1854 | in F<lib/>, so here's some opportunity for some patching. |
f7e1e956 MS |
1855 | |
1856 | You must be triply conscious of cross-platform concerns. This usually | |
1857 | boils down to using File::Spec and avoiding things like C<fork()> and | |
1858 | C<system()> unless absolutely necessary. | |
1859 | ||
e018f8be JH |
1860 | =head2 Special Make Test Targets |
1861 | ||
1862 | There are various special make targets that can be used to test Perl | |
1863 | slightly differently than the standard "test" target. Not all them | |
1864 | are expected to give a 100% success rate. Many of them have several | |
7205a85d YO |
1865 | aliases, and many of them are not available on certain operating |
1866 | systems. | |
e018f8be JH |
1867 | |
1868 | =over 4 | |
1869 | ||
1870 | =item coretest | |
1871 | ||
7d7d5695 | 1872 | Run F<perl> on all core tests (F<t/*> and F<lib/[a-z]*> pragma tests). |
e018f8be | 1873 | |
7205a85d YO |
1874 | (Not available on Win32) |
1875 | ||
e018f8be JH |
1876 | =item test.deparse |
1877 | ||
b26492ee RGS |
1878 | Run all the tests through B::Deparse. Not all tests will succeed. |
1879 | ||
7205a85d YO |
1880 | (Not available on Win32) |
1881 | ||
b26492ee RGS |
1882 | =item test.taintwarn |
1883 | ||
1884 | Run all tests with the B<-t> command-line switch. Not all tests | |
1885 | are expected to succeed (until they're specifically fixed, of course). | |
e018f8be | 1886 | |
7205a85d YO |
1887 | (Not available on Win32) |
1888 | ||
e018f8be JH |
1889 | =item minitest |
1890 | ||
1891 | Run F<miniperl> on F<t/base>, F<t/comp>, F<t/cmd>, F<t/run>, F<t/io>, | |
8cebccf4 | 1892 | F<t/op>, F<t/uni> and F<t/mro> tests. |
e018f8be | 1893 | |
7a834142 JH |
1894 | =item test.valgrind check.valgrind utest.valgrind ucheck.valgrind |
1895 | ||
1896 | (Only in Linux) Run all the tests using the memory leak + naughty | |
1897 | memory access tool "valgrind". The log files will be named | |
1898 | F<testname.valgrind>. | |
1899 | ||
e018f8be JH |
1900 | =item test.third check.third utest.third ucheck.third |
1901 | ||
1902 | (Only in Tru64) Run all the tests using the memory leak + naughty | |
1903 | memory access tool "Third Degree". The log files will be named | |
60a57c1c | 1904 | F<perl.3log.testname>. |
e018f8be JH |
1905 | |
1906 | =item test.torture torturetest | |
1907 | ||
1908 | Run all the usual tests and some extra tests. As of Perl 5.8.0 the | |
244d9cb7 | 1909 | only extra tests are Abigail's JAPHs, F<t/japh/abigail.t>. |
e018f8be JH |
1910 | |
1911 | You can also run the torture test with F<t/harness> by giving | |
1912 | C<-torture> argument to F<t/harness>. | |
1913 | ||
1914 | =item utest ucheck test.utf8 check.utf8 | |
1915 | ||
1916 | Run all the tests with -Mutf8. Not all tests will succeed. | |
1917 | ||
7205a85d YO |
1918 | (Not available on Win32) |
1919 | ||
cc0710ff RGS |
1920 | =item minitest.utf16 test.utf16 |
1921 | ||
1922 | Runs the tests with UTF-16 encoded scripts, encoded with different | |
1923 | versions of this encoding. | |
1924 | ||
1925 | C<make utest.utf16> runs the test suite with a combination of C<-utf8> and | |
1926 | C<-utf16> arguments to F<t/TEST>. | |
1927 | ||
7205a85d YO |
1928 | (Not available on Win32) |
1929 | ||
244d9cb7 RGS |
1930 | =item test_harness |
1931 | ||
1932 | Run the test suite with the F<t/harness> controlling program, instead of | |
1933 | F<t/TEST>. F<t/harness> is more sophisticated, and uses the | |
1934 | L<Test::Harness> module, thus using this test target supposes that perl | |
1935 | mostly works. The main advantage for our purposes is that it prints a | |
00bf5cd9 RGS |
1936 | detailed summary of failed tests at the end. Also, unlike F<t/TEST>, it |
1937 | doesn't redirect stderr to stdout. | |
244d9cb7 | 1938 | |
7205a85d YO |
1939 | Note that under Win32 F<t/harness> is always used instead of F<t/TEST>, so |
1940 | there is no special "test_harness" target. | |
1941 | ||
1942 | Under Win32's "test" target you may use the TEST_SWITCHES and TEST_FILES | |
1943 | environment variables to control the behaviour of F<t/harness>. This means | |
1944 | you can say | |
1945 | ||
1946 | nmake test TEST_FILES="op/*.t" | |
1947 | nmake test TEST_SWITCHES="-torture" TEST_FILES="op/*.t" | |
1948 | ||
a75f557c JV |
1949 | =item Parallel tests |
1950 | ||
1951 | The core distribution can now run its regression tests in parallel on | |
1952 | Unix-like platforms. Instead of running C<make test>, set C<TEST_JOBS> in | |
1953 | your environment to the number of tests to run in parallel, and run | |
1954 | C<make test_harness>. On a Bourne-like shell, this can be done as | |
1955 | ||
1956 | TEST_JOBS=3 make test_harness # Run 3 tests in parallel | |
1957 | ||
1958 | An environment variable is used, rather than parallel make itself, because | |
1959 | L<TAP::Harness> needs to be able to schedule individual non-conflicting test | |
1960 | scripts itself, and there is no standard interface to C<make> utilities to | |
1961 | interact with their job schedulers. | |
1962 | ||
1963 | Note that currently some test scripts may fail when run in parallel (most | |
1964 | notably C<ext/IO/t/io_dir.t>). If necessary run just the failing scripts | |
1965 | again sequentially and see if the failures go away. | |
7205a85d YO |
1966 | =item test-notty test_notty |
1967 | ||
1968 | Sets PERL_SKIP_TTY_TEST to true before running normal test. | |
1969 | ||
244d9cb7 RGS |
1970 | =back |
1971 | ||
1972 | =head2 Running tests by hand | |
1973 | ||
1974 | You can run part of the test suite by hand by using one the following | |
1975 | commands from the F<t/> directory : | |
1976 | ||
1977 | ./perl -I../lib TEST list-of-.t-files | |
1978 | ||
1979 | or | |
1980 | ||
1981 | ./perl -I../lib harness list-of-.t-files | |
1982 | ||
1983 | (if you don't specify test scripts, the whole test suite will be run.) | |
1984 | ||
7205a85d YO |
1985 | =head3 Using t/harness for testing |
1986 | ||
1987 | If you use C<harness> for testing you have several command line options | |
1988 | available to you. The arguments are as follows, and are in the order | |
1989 | that they must appear if used together. | |
1990 | ||
1991 | harness -v -torture -re=pattern LIST OF FILES TO TEST | |
1992 | harness -v -torture -re LIST OF PATTERNS TO MATCH | |
1993 | ||
1994 | If C<LIST OF FILES TO TEST> is omitted the file list is obtained from | |
07aa3531 | 1995 | the manifest. The file list may include shell wildcards which will be |
7205a85d YO |
1996 | expanded out. |
1997 | ||
1998 | =over 4 | |
1999 | ||
2000 | =item -v | |
2001 | ||
07aa3531 | 2002 | Run the tests under verbose mode so you can see what tests were run, |
8550bf48 | 2003 | and debug output. |
7205a85d YO |
2004 | |
2005 | =item -torture | |
2006 | ||
2007 | Run the torture tests as well as the normal set. | |
2008 | ||
2009 | =item -re=PATTERN | |
2010 | ||
2011 | Filter the file list so that all the test files run match PATTERN. | |
2012 | Note that this form is distinct from the B<-re LIST OF PATTERNS> form below | |
2013 | in that it allows the file list to be provided as well. | |
2014 | ||
2015 | =item -re LIST OF PATTERNS | |
2016 | ||
07aa3531 | 2017 | Filter the file list so that all the test files run match |
7205a85d YO |
2018 | /(LIST|OF|PATTERNS)/. Note that with this form the patterns |
2019 | are joined by '|' and you cannot supply a list of files, instead | |
2020 | the test files are obtained from the MANIFEST. | |
2021 | ||
2022 | =back | |
2023 | ||
244d9cb7 RGS |
2024 | You can run an individual test by a command similar to |
2025 | ||
2026 | ./perl -I../lib patho/to/foo.t | |
2027 | ||
2028 | except that the harnesses set up some environment variables that may | |
2029 | affect the execution of the test : | |
2030 | ||
07aa3531 | 2031 | =over 4 |
244d9cb7 RGS |
2032 | |
2033 | =item PERL_CORE=1 | |
2034 | ||
2035 | indicates that we're running this test part of the perl core test suite. | |
2036 | This is useful for modules that have a dual life on CPAN. | |
2037 | ||
2038 | =item PERL_DESTRUCT_LEVEL=2 | |
2039 | ||
2040 | is set to 2 if it isn't set already (see L</PERL_DESTRUCT_LEVEL>) | |
2041 | ||
2042 | =item PERL | |
2043 | ||
2044 | (used only by F<t/TEST>) if set, overrides the path to the perl executable | |
2045 | that should be used to run the tests (the default being F<./perl>). | |
2046 | ||
2047 | =item PERL_SKIP_TTY_TEST | |
2048 | ||
2049 | if set, tells to skip the tests that need a terminal. It's actually set | |
2050 | automatically by the Makefile, but can also be forced artificially by | |
2051 | running 'make test_notty'. | |
2052 | ||
e018f8be | 2053 | =back |
f7e1e956 | 2054 | |
7cd58830 RGS |
2055 | =head3 Other environment variables that may influence tests |
2056 | ||
2057 | =over 4 | |
2058 | ||
2059 | =item PERL_TEST_Net_Ping | |
2060 | ||
2061 | Setting this variable runs all the Net::Ping modules tests, | |
2062 | otherwise some tests that interact with the outside world are skipped. | |
2063 | See L<perl58delta>. | |
2064 | ||
2065 | =item PERL_TEST_NOVREXX | |
2066 | ||
2067 | Setting this variable skips the vrexx.t tests for OS2::REXX. | |
2068 | ||
2069 | =item PERL_TEST_NUMCONVERTS | |
2070 | ||
2071 | This sets a variable in op/numconvert.t. | |
2072 | ||
2073 | =back | |
2074 | ||
2075 | See also the documentation for the Test and Test::Harness modules, | |
2076 | for more environment variables that affect testing. | |
2077 | ||
d7889f52 JH |
2078 | =head2 Common problems when patching Perl source code |
2079 | ||
2080 | Perl source plays by ANSI C89 rules: no C99 (or C++) extensions. In | |
2081 | some cases we have to take pre-ANSI requirements into consideration. | |
2082 | You don't care about some particular platform having broken Perl? | |
2083 | I hear there is still a strong demand for J2EE programmers. | |
2084 | ||
2085 | =head2 Perl environment problems | |
2086 | ||
2087 | =over 4 | |
2088 | ||
2089 | =item * | |
2090 | ||
2091 | Not compiling with threading | |
2092 | ||
2093 | Compiling with threading (-Duseithreads) completely rewrites | |
2094 | the function prototypes of Perl. You better try your changes | |
0bec6c03 | 2095 | with that. Related to this is the difference between "Perl_-less" |
d7889f52 JH |
2096 | and "Perl_-ly" APIs, for example: |
2097 | ||
2098 | Perl_sv_setiv(aTHX_ ...); | |
2099 | sv_setiv(...); | |
2100 | ||
ee9468a2 RGS |
2101 | The first one explicitly passes in the context, which is needed for e.g. |
2102 | threaded builds. The second one does that implicitly; do not get them | |
def4ed7d JH |
2103 | mixed. If you are not passing in a aTHX_, you will need to do a dTHX |
2104 | (or a dVAR) as the first thing in the function. | |
d7889f52 JH |
2105 | |
2106 | See L<perlguts/"How multiple interpreters and concurrency are supported"> | |
2107 | for further discussion about context. | |
2108 | ||
2109 | =item * | |
2110 | ||
2111 | Not compiling with -DDEBUGGING | |
2112 | ||
2113 | The DEBUGGING define exposes more code to the compiler, | |
0bec6c03 | 2114 | therefore more ways for things to go wrong. You should try it. |
d7889f52 JH |
2115 | |
2116 | =item * | |
2117 | ||
ee9468a2 RGS |
2118 | Introducing (non-read-only) globals |
2119 | ||
2120 | Do not introduce any modifiable globals, truly global or file static. | |
bc028b6b JH |
2121 | They are bad form and complicate multithreading and other forms of |
2122 | concurrency. The right way is to introduce them as new interpreter | |
2123 | variables, see F<intrpvar.h> (at the very end for binary compatibility). | |
ee9468a2 RGS |
2124 | |
2125 | Introducing read-only (const) globals is okay, as long as you verify | |
2126 | with e.g. C<nm libperl.a|egrep -v ' [TURtr] '> (if your C<nm> has | |
2127 | BSD-style output) that the data you added really is read-only. | |
2128 | (If it is, it shouldn't show up in the output of that command.) | |
2129 | ||
2130 | If you want to have static strings, make them constant: | |
2131 | ||
2132 | static const char etc[] = "..."; | |
2133 | ||
bc028b6b | 2134 | If you want to have arrays of constant strings, note carefully |
ee9468a2 RGS |
2135 | the right combination of C<const>s: |
2136 | ||
2137 | static const char * const yippee[] = | |
2138 | {"hi", "ho", "silver"}; | |
2139 | ||
bc028b6b JH |
2140 | There is a way to completely hide any modifiable globals (they are all |
2141 | moved to heap), the compilation setting C<-DPERL_GLOBAL_STRUCT_PRIVATE>. | |
2142 | It is not normally used, but can be used for testing, read more | |
def4ed7d | 2143 | about it in L<perlguts/"Background and PERL_IMPLICIT_CONTEXT">. |
bc028b6b | 2144 | |
ee9468a2 RGS |
2145 | =item * |
2146 | ||
d7889f52 JH |
2147 | Not exporting your new function |
2148 | ||
2149 | Some platforms (Win32, AIX, VMS, OS/2, to name a few) require any | |
2150 | function that is part of the public API (the shared Perl library) | |
2151 | to be explicitly marked as exported. See the discussion about | |
2152 | F<embed.pl> in L<perlguts>. | |
2153 | ||
2154 | =item * | |
2155 | ||
2156 | Exporting your new function | |
2157 | ||
2158 | The new shiny result of either genuine new functionality or your | |
2159 | arduous refactoring is now ready and correctly exported. So what | |
def4ed7d | 2160 | could possibly go wrong? |
d7889f52 JH |
2161 | |
2162 | Maybe simply that your function did not need to be exported in the | |
2163 | first place. Perl has a long and not so glorious history of exporting | |
2164 | functions that it should not have. | |
2165 | ||
2166 | If the function is used only inside one source code file, make it | |
2167 | static. See the discussion about F<embed.pl> in L<perlguts>. | |
2168 | ||
2169 | If the function is used across several files, but intended only for | |
2170 | Perl's internal use (and this should be the common case), do not | |
2171 | export it to the public API. See the discussion about F<embed.pl> | |
2172 | in L<perlguts>. | |
2173 | ||
2174 | =back | |
2175 | ||
5b38b9cd | 2176 | =head2 Portability problems |
d7889f52 JH |
2177 | |
2178 | The following are common causes of compilation and/or execution | |
2179 | failures, not common to Perl as such. The C FAQ is good bedtime | |
0bec6c03 | 2180 | reading. Please test your changes with as many C compilers and |
ac036724 | 2181 | platforms as possible; we will, anyway, and it's nice to save |
0bec6c03 JH |
2182 | oneself from public embarrassment. |
2183 | ||
9aaf14db RGS |
2184 | If using gcc, you can add the C<-std=c89> option which will hopefully |
2185 | catch most of these unportabilities. (However it might also catch | |
2186 | incompatibilities in your system's header files.) | |
d1307786 | 2187 | |
a8e98a71 JH |
2188 | Use the Configure C<-Dgccansipedantic> flag to enable the gcc |
2189 | C<-ansi -pedantic> flags which enforce stricter ANSI rules. | |
2190 | ||
def4ed7d JH |
2191 | If using the C<gcc -Wall> note that not all the possible warnings |
2192 | (like C<-Wunitialized>) are given unless you also compile with C<-O>. | |
2193 | ||
2194 | Note that if using gcc, starting from Perl 5.9.5 the Perl core source | |
2195 | code files (the ones at the top level of the source code distribution, | |
2196 | but not e.g. the extensions under ext/) are automatically compiled | |
2197 | with as many as possible of the C<-std=c89>, C<-ansi>, C<-pedantic>, | |
2198 | and a selection of C<-W> flags (see cflags.SH). | |
27565cb6 | 2199 | |
0bec6c03 | 2200 | Also study L<perlport> carefully to avoid any bad assumptions |
def4ed7d | 2201 | about the operating system, filesystems, and so forth. |
0bec6c03 | 2202 | |
606fd33d | 2203 | You may once in a while try a "make microperl" to see whether we |
63796a85 | 2204 | can still compile Perl with just the bare minimum of interfaces. |
606fd33d | 2205 | (See README.micro.) |
ee9468a2 | 2206 | |
0bec6c03 | 2207 | Do not assume an operating system indicates a certain compiler. |
d7889f52 JH |
2208 | |
2209 | =over 4 | |
2210 | ||
2211 | =item * | |
2212 | ||
2213 | Casting pointers to integers or casting integers to pointers | |
2214 | ||
2215 | void castaway(U8* p) | |
2216 | { | |
2217 | IV i = p; | |
2218 | ||
2219 | or | |
2220 | ||
2221 | void castaway(U8* p) | |
2222 | { | |
2223 | IV i = (IV)p; | |
2224 | ||
ee9468a2 | 2225 | Both are bad, and broken, and unportable. Use the PTR2IV() |
d7889f52 JH |
2226 | macro that does it right. (Likewise, there are PTR2UV(), PTR2NV(), |
2227 | INT2PTR(), and NUM2PTR().) | |
2228 | ||
2229 | =item * | |
2230 | ||
0bec6c03 JH |
2231 | Casting between data function pointers and data pointers |
2232 | ||
d7889f52 JH |
2233 | Technically speaking casting between function pointers and data |
2234 | pointers is unportable and undefined, but practically speaking | |
2235 | it seems to work, but you should use the FPTR2DPTR() and DPTR2FPTR() | |
0bec6c03 | 2236 | macros. Sometimes you can also play games with unions. |
d7889f52 JH |
2237 | |
2238 | =item * | |
2239 | ||
2240 | Assuming sizeof(int) == sizeof(long) | |
2241 | ||
2242 | There are platforms where longs are 64 bits, and platforms where ints | |
2243 | are 64 bits, and while we are out to shock you, even platforms where | |
2244 | shorts are 64 bits. This is all legal according to the C standard. | |
2245 | (In other words, "long long" is not a portable way to specify 64 bits, | |
2246 | and "long long" is not even guaranteed to be any wider than "long".) | |
63796a85 JH |
2247 | |
2248 | Instead, use the definitions IV, UV, IVSIZE, I32SIZE, and so forth. | |
2249 | Avoid things like I32 because they are B<not> guaranteed to be | |
2250 | I<exactly> 32 bits, they are I<at least> 32 bits, nor are they | |
2251 | guaranteed to be B<int> or B<long>. If you really explicitly need | |
2252 | 64-bit variables, use I64 and U64, but only if guarded by HAS_QUAD. | |
d7889f52 JH |
2253 | |
2254 | =item * | |
2255 | ||
2256 | Assuming one can dereference any type of pointer for any type of data | |
2257 | ||
2258 | char *p = ...; | |
def4ed7d | 2259 | long pony = *p; /* BAD */ |
d7889f52 JH |
2260 | |
2261 | Many platforms, quite rightly so, will give you a core dump instead | |
2262 | of a pony if the p happens not be correctly aligned. | |
2263 | ||
2264 | =item * | |
2265 | ||
2266 | Lvalue casts | |
2267 | ||
def4ed7d | 2268 | (int)*p = ...; /* BAD */ |
d7889f52 JH |
2269 | |
2270 | Simply not portable. Get your lvalue to be of the right type, | |
27565cb6 JH |
2271 | or maybe use temporary variables, or dirty tricks with unions. |
2272 | ||
2273 | =item * | |
2274 | ||
606fd33d JH |
2275 | Assume B<anything> about structs (especially the ones you |
2276 | don't control, like the ones coming from the system headers) | |
27565cb6 JH |
2277 | |
2278 | =over 8 | |
2279 | ||
2280 | =item * | |
2281 | ||
2282 | That a certain field exists in a struct | |
2283 | ||
2284 | =item * | |
2285 | ||
902821cc | 2286 | That no other fields exist besides the ones you know of |
27565cb6 JH |
2287 | |
2288 | =item * | |
2289 | ||
606fd33d | 2290 | That a field is of certain signedness, sizeof, or type |
27565cb6 JH |
2291 | |
2292 | =item * | |
2293 | ||
2294 | That the fields are in a certain order | |
2295 | ||
606fd33d JH |
2296 | =over 8 |
2297 | ||
27565cb6 JH |
2298 | =item * |
2299 | ||
606fd33d JH |
2300 | While C guarantees the ordering specified in the struct definition, |
2301 | between different platforms the definitions might differ | |
2302 | ||
2303 | =back | |
27565cb6 JH |
2304 | |
2305 | =item * | |
2306 | ||
606fd33d JH |
2307 | That the sizeof(struct) or the alignments are the same everywhere |
2308 | ||
2309 | =over 8 | |
27565cb6 JH |
2310 | |
2311 | =item * | |
2312 | ||
606fd33d JH |
2313 | There might be padding bytes between the fields to align the fields - |
2314 | the bytes can be anything | |
2315 | ||
2316 | =item * | |
2317 | ||
2318 | Structs are required to be aligned to the maximum alignment required | |
2319 | by the fields - which for native types is for usually equivalent to | |
2320 | sizeof() of the field | |
2321 | ||
2322 | =back | |
27565cb6 JH |
2323 | |
2324 | =back | |
d7889f52 JH |
2325 | |
2326 | =item * | |
2327 | ||
2bbc8d55 SP |
2328 | Assuming the character set is ASCIIish |
2329 | ||
2330 | Perl can compile and run under EBCDIC platforms. See L<perlebcdic>. | |
2331 | This is transparent for the most part, but because the character sets | |
2332 | differ, you shouldn't use numeric (decimal, octal, nor hex) constants | |
2333 | to refer to characters. You can safely say 'A', but not 0x41. | |
2334 | You can safely say '\n', but not \012. | |
2335 | If a character doesn't have a trivial input form, you can | |
2336 | create a #define for it in both C<utfebcdic.h> and C<utf8.h>, so that | |
2337 | it resolves to different values depending on the character set being used. | |
2338 | (There are three different EBCDIC character sets defined in C<utfebcdic.h>, | |
2339 | so it might be best to insert the #define three times in that file.) | |
2340 | ||
2341 | Also, the range 'A' - 'Z' in ASCII is an unbroken sequence of 26 upper case | |
2342 | alphabetic characters. That is not true in EBCDIC. Nor for 'a' to 'z'. | |
2343 | But '0' - '9' is an unbroken range in both systems. Don't assume anything | |
2344 | about other ranges. | |
2345 | ||
2346 | Many of the comments in the existing code ignore the possibility of EBCDIC, | |
2347 | and may be wrong therefore, even if the code works. | |
2348 | This is actually a tribute to the successful transparent insertion of being | |
fe749c9a | 2349 | able to handle EBCDIC without having to change pre-existing code. |
2bbc8d55 SP |
2350 | |
2351 | UTF-8 and UTF-EBCDIC are two different encodings used to represent Unicode | |
2352 | code points as sequences of bytes. Macros | |
2353 | with the same names (but different definitions) | |
2354 | in C<utf8.h> and C<utfebcdic.h> | |
fe749c9a KW |
2355 | are used to allow the calling code to think that there is only one such |
2356 | encoding. | |
2357 | This is almost always referred to as C<utf8>, but it means the EBCDIC version | |
2358 | as well. Again, comments in the code may well be wrong even if the code itself | |
2359 | is right. | |
2bbc8d55 SP |
2360 | For example, the concept of C<invariant characters> differs between ASCII and |
2361 | EBCDIC. | |
2362 | On ASCII platforms, only characters that do not have the high-order | |
2363 | bit set (i.e. whose ordinals are strict ASCII, 0 - 127) | |
2364 | are invariant, and the documentation and comments in the code | |
2365 | may assume that, | |
2366 | often referring to something like, say, C<hibit>. | |
2367 | The situation differs and is not so simple on EBCDIC machines, but as long as | |
2368 | the code itself uses the C<NATIVE_IS_INVARIANT()> macro appropriately, it | |
2369 | works, even if the comments are wrong. | |
2370 | ||
2371 | =item * | |
2372 | ||
2373 | Assuming the character set is just ASCII | |
2374 | ||
2375 | ASCII is a 7 bit encoding, but bytes have 8 bits in them. The 128 extra | |
2376 | characters have different meanings depending on the locale. Absent a locale, | |
2377 | currently these extra characters are generally considered to be unassigned, | |
2378 | and this has presented some problems. | |
e1b711da | 2379 | This is being changed starting in 5.12 so that these characters will |
2bbc8d55 SP |
2380 | be considered to be Latin-1 (ISO-8859-1). |
2381 | ||
2382 | =item * | |
2383 | ||
0bec6c03 JH |
2384 | Mixing #define and #ifdef |
2385 | ||
2386 | #define BURGLE(x) ... \ | |
def4ed7d | 2387 | #ifdef BURGLE_OLD_STYLE /* BAD */ |
0bec6c03 JH |
2388 | ... do it the old way ... \ |
2389 | #else | |
2390 | ... do it the new way ... \ | |
2391 | #endif | |
2392 | ||
ee9468a2 RGS |
2393 | You cannot portably "stack" cpp directives. For example in the above |
2394 | you need two separate BURGLE() #defines, one for each #ifdef branch. | |
2395 | ||
2396 | =item * | |
2397 | ||
2bbc8d55 | 2398 | Adding non-comment stuff after #endif or #else |
ee9468a2 RGS |
2399 | |
2400 | #ifdef SNOSH | |
2401 | ... | |
def4ed7d | 2402 | #else !SNOSH /* BAD */ |
ee9468a2 | 2403 | ... |
def4ed7d | 2404 | #endif SNOSH /* BAD */ |
ee9468a2 | 2405 | |
def4ed7d JH |
2406 | The #endif and #else cannot portably have anything non-comment after |
2407 | them. If you want to document what is going (which is a good idea | |
2408 | especially if the branches are long), use (C) comments: | |
ee9468a2 RGS |
2409 | |
2410 | #ifdef SNOSH | |
2411 | ... | |
2412 | #else /* !SNOSH */ | |
2413 | ... | |
2414 | #endif /* SNOSH */ | |
2415 | ||
2416 | The gcc option C<-Wendif-labels> warns about the bad variant | |
2417 | (by default on starting from Perl 5.9.4). | |
0bec6c03 JH |
2418 | |
2419 | =item * | |
2420 | ||
27565cb6 JH |
2421 | Having a comma after the last element of an enum list |
2422 | ||
2423 | enum color { | |
2424 | CERULEAN, | |
2425 | CHARTREUSE, | |
def4ed7d | 2426 | CINNABAR, /* BAD */ |
27565cb6 JH |
2427 | }; |
2428 | ||
2429 | is not portable. Leave out the last comma. | |
2430 | ||
2431 | Also note that whether enums are implicitly morphable to ints | |
2432 | varies between compilers, you might need to (int). | |
2433 | ||
2434 | =item * | |
2435 | ||
d7889f52 JH |
2436 | Using //-comments |
2437 | ||
def4ed7d | 2438 | // This function bamfoodles the zorklator. /* BAD */ |
d7889f52 JH |
2439 | |
2440 | That is C99 or C++. Perl is C89. Using the //-comments is silently | |
0bec6c03 JH |
2441 | allowed by many C compilers but cranking up the ANSI C89 strictness |
2442 | (which we like to do) causes the compilation to fail. | |
d7889f52 JH |
2443 | |
2444 | =item * | |
2445 | ||
2446 | Mixing declarations and code | |
2447 | ||
2448 | void zorklator() | |
2449 | { | |
2450 | int n = 3; | |
def4ed7d | 2451 | set_zorkmids(n); /* BAD */ |
d7889f52 JH |
2452 | int q = 4; |
2453 | ||
0bec6c03 JH |
2454 | That is C99 or C++. Some C compilers allow that, but you shouldn't. |
2455 | ||
63796a85 JH |
2456 | The gcc option C<-Wdeclaration-after-statements> scans for such problems |
2457 | (by default on starting from Perl 5.9.4). | |
2458 | ||
0bec6c03 JH |
2459 | =item * |
2460 | ||
2461 | Introducing variables inside for() | |
2462 | ||
def4ed7d | 2463 | for(int i = ...; ...; ...) { /* BAD */ |
0bec6c03 JH |
2464 | |
2465 | That is C99 or C++. While it would indeed be awfully nice to have that | |
2466 | also in C89, to limit the scope of the loop variable, alas, we cannot. | |
d7889f52 JH |
2467 | |
2468 | =item * | |
2469 | ||
2470 | Mixing signed char pointers with unsigned char pointers | |
2471 | ||
2472 | int foo(char *s) { ... } | |
2473 | ... | |
2474 | unsigned char *t = ...; /* Or U8* t = ... */ | |
def4ed7d | 2475 | foo(t); /* BAD */ |
d7889f52 JH |
2476 | |
2477 | While this is legal practice, it is certainly dubious, and downright | |
2478 | fatal in at least one platform: for example VMS cc considers this a | |
def4ed7d JH |
2479 | fatal error. One cause for people often making this mistake is that a |
2480 | "naked char" and therefore dereferencing a "naked char pointer" have | |
2481 | an undefined signedness: it depends on the compiler and the flags of | |
2482 | the compiler and the underlying platform whether the result is signed | |
2483 | or unsigned. For this very same reason using a 'char' as an array | |
2484 | index is bad. | |
d7889f52 JH |
2485 | |
2486 | =item * | |
2487 | ||
2488 | Macros that have string constants and their arguments as substrings of | |
2489 | the string constants | |
2490 | ||
def4ed7d | 2491 | #define FOO(n) printf("number = %d\n", n) /* BAD */ |
d7889f52 JH |
2492 | FOO(10); |
2493 | ||
2494 | Pre-ANSI semantics for that was equivalent to | |
2495 | ||
2496 | printf("10umber = %d\10"); | |
2497 | ||
0bec6c03 JH |
2498 | which is probably not what you were expecting. Unfortunately at least |
2499 | one reasonably common and modern C compiler does "real backward | |
63796a85 | 2500 | compatibility" here, in AIX that is what still happens even though the |
0bec6c03 JH |
2501 | rest of the AIX compiler is very happily C89. |
2502 | ||
2503 | =item * | |
2504 | ||
ee9468a2 RGS |
2505 | Using printf formats for non-basic C types |
2506 | ||
2507 | IV i = ...; | |
def4ed7d | 2508 | printf("i = %d\n", i); /* BAD */ |
ee9468a2 RGS |
2509 | |
2510 | While this might by accident work in some platform (where IV happens | |
2511 | to be an C<int>), in general it cannot. IV might be something larger. | |
2512 | Even worse the situation is with more specific types (defined by Perl's | |
2513 | configuration step in F<config.h>): | |
2514 | ||
2515 | Uid_t who = ...; | |
def4ed7d | 2516 | printf("who = %d\n", who); /* BAD */ |
ee9468a2 RGS |
2517 | |
2518 | The problem here is that Uid_t might be not only not C<int>-wide | |
2519 | but it might also be unsigned, in which case large uids would be | |
2520 | printed as negative values. | |
2521 | ||
2522 | There is no simple solution to this because of printf()'s limited | |
2523 | intelligence, but for many types the right format is available as | |
2524 | with either 'f' or '_f' suffix, for example: | |
2525 | ||
2526 | IVdf /* IV in decimal */ | |
2527 | UVxf /* UV is hexadecimal */ | |
2528 | ||
2529 | printf("i = %"IVdf"\n", i); /* The IVdf is a string constant. */ | |
2530 | ||
2531 | Uid_t_f /* Uid_t in decimal */ | |
2532 | ||
2533 | printf("who = %"Uid_t_f"\n", who); | |
2534 | ||
63796a85 JH |
2535 | Or you can try casting to a "wide enough" type: |
2536 | ||
2537 | printf("i = %"IVdf"\n", (IV)something_very_small_and_signed); | |
2538 | ||
2539 | Also remember that the C<%p> format really does require a void pointer: | |
2540 | ||
2541 | U8* p = ...; | |
2542 | printf("p = %p\n", (void*)p); | |
2543 | ||
ee9468a2 RGS |
2544 | The gcc option C<-Wformat> scans for such problems. |
2545 | ||
2546 | =item * | |
2547 | ||
0bec6c03 JH |
2548 | Blindly using variadic macros |
2549 | ||
63796a85 JH |
2550 | gcc has had them for a while with its own syntax, and C99 brought |
2551 | them with a standardized syntax. Don't use the former, and use | |
2552 | the latter only if the HAS_C99_VARIADIC_MACROS is defined. | |
0bec6c03 JH |
2553 | |
2554 | =item * | |
2555 | ||
2556 | Blindly passing va_list | |
2557 | ||
2558 | Not all platforms support passing va_list to further varargs (stdarg) | |
2559 | functions. The right thing to do is to copy the va_list using the | |
2560 | Perl_va_copy() if the NEED_VA_COPY is defined. | |
d7889f52 | 2561 | |
ee9468a2 RGS |
2562 | =item * |
2563 | ||
e5afc1ae | 2564 | Using gcc statement expressions |
63796a85 | 2565 | |
def4ed7d | 2566 | val = ({...;...;...}); /* BAD */ |
63796a85 | 2567 | |
def4ed7d JH |
2568 | While a nice extension, it's not portable. The Perl code does |
2569 | admittedly use them if available to gain some extra speed | |
2570 | (essentially as a funky form of inlining), but you shouldn't. | |
63796a85 JH |
2571 | |
2572 | =item * | |
2573 | ||
2bbc8d55 | 2574 | Binding together several statements in a macro |
63796a85 JH |
2575 | |
2576 | Use the macros STMT_START and STMT_END. | |
2577 | ||
2578 | STMT_START { | |
2579 | ... | |
2580 | } STMT_END | |
2581 | ||
2582 | =item * | |
2583 | ||
ee9468a2 RGS |
2584 | Testing for operating systems or versions when should be testing for features |
2585 | ||
def4ed7d | 2586 | #ifdef __FOONIX__ /* BAD */ |
ee9468a2 RGS |
2587 | foo = quux(); |
2588 | #endif | |
2589 | ||
2590 | Unless you know with 100% certainty that quux() is only ever available | |
2591 | for the "Foonix" operating system B<and> that is available B<and> | |
2592 | correctly working for B<all> past, present, B<and> future versions of | |
2593 | "Foonix", the above is very wrong. This is more correct (though still | |
2594 | not perfect, because the below is a compile-time check): | |
2595 | ||
2596 | #ifdef HAS_QUUX | |
2597 | foo = quux(); | |
2598 | #endif | |
2599 | ||
def4ed7d | 2600 | How does the HAS_QUUX become defined where it needs to be? Well, if |
e1020413 | 2601 | Foonix happens to be Unixy enough to be able to run the Configure |
ee9468a2 RGS |
2602 | script, and Configure has been taught about detecting and testing |
2603 | quux(), the HAS_QUUX will be correctly defined. In other platforms, | |
2604 | the corresponding configuration step will hopefully do the same. | |
2605 | ||
2606 | In a pinch, if you cannot wait for Configure to be educated, | |
2607 | or if you have a good hunch of where quux() might be available, | |
2608 | you can temporarily try the following: | |
2609 | ||
2610 | #if (defined(__FOONIX__) || defined(__BARNIX__)) | |
2611 | # define HAS_QUUX | |
2612 | #endif | |
2613 | ||
2614 | ... | |
2615 | ||
2616 | #ifdef HAS_QUUX | |
2617 | foo = quux(); | |
2618 | #endif | |
2619 | ||
2620 | But in any case, try to keep the features and operating systems separate. | |
2621 | ||
d7889f52 JH |
2622 | =back |
2623 | ||
ad7244db JH |
2624 | =head2 Problematic System Interfaces |
2625 | ||
2626 | =over 4 | |
2627 | ||
2628 | =item * | |
2629 | ||
353c6505 | 2630 | malloc(0), realloc(0), calloc(0, 0) are non-portable. To be portable |
ad7244db JH |
2631 | allocate at least one byte. (In general you should rarely need to |
2632 | work at this low level, but instead use the various malloc wrappers.) | |
2633 | ||
2634 | =item * | |
2635 | ||
2636 | snprintf() - the return type is unportable. Use my_snprintf() instead. | |
2637 | ||
2638 | =back | |
2639 | ||
d7889f52 JH |
2640 | =head2 Security problems |
2641 | ||
2642 | Last but not least, here are various tips for safer coding. | |
2643 | ||
2644 | =over 4 | |
2645 | ||
2646 | =item * | |
2647 | ||
2648 | Do not use gets() | |
2649 | ||
2650 | Or we will publicly ridicule you. Seriously. | |
2651 | ||
2652 | =item * | |
2653 | ||
d1307786 | 2654 | Do not use strcpy() or strcat() or strncpy() or strncat() |
d7889f52 | 2655 | |
d1307786 JH |
2656 | Use my_strlcpy() and my_strlcat() instead: they either use the native |
2657 | implementation, or Perl's own implementation (borrowed from the public | |
2658 | domain implementation of INN). | |
d7889f52 JH |
2659 | |
2660 | =item * | |
2661 | ||
2662 | Do not use sprintf() or vsprintf() | |
2663 | ||
0bec6c03 | 2664 | If you really want just plain byte strings, use my_snprintf() |
64d9b66b | 2665 | and my_vsnprintf() instead, which will try to use snprintf() and |
0bec6c03 JH |
2666 | vsnprintf() if those safer APIs are available. If you want something |
2667 | fancier than a plain byte string, use SVs and Perl_sv_catpvf(). | |
d7889f52 JH |
2668 | |
2669 | =back | |
2670 | ||
902b9dbf MLF |
2671 | =head1 EXTERNAL TOOLS FOR DEBUGGING PERL |
2672 | ||
2673 | Sometimes it helps to use external tools while debugging and | |
2674 | testing Perl. This section tries to guide you through using | |
2675 | some common testing and debugging tools with Perl. This is | |
2676 | meant as a guide to interfacing these tools with Perl, not | |
2677 | as any kind of guide to the use of the tools themselves. | |
2678 | ||
a958818a JH |
2679 | B<NOTE 1>: Running under memory debuggers such as Purify, valgrind, or |
2680 | Third Degree greatly slows down the execution: seconds become minutes, | |
2681 | minutes become hours. For example as of Perl 5.8.1, the | |
2682 | ext/Encode/t/Unicode.t takes extraordinarily long to complete under | |
2683 | e.g. Purify, Third Degree, and valgrind. Under valgrind it takes more | |
ac036724 | 2684 | than six hours, even on a snappy computer. The said test must be |
a958818a JH |
2685 | doing something that is quite unfriendly for memory debuggers. If you |
2686 | don't feel like waiting, that you can simply kill away the perl | |
2687 | process. | |
2688 | ||
2689 | B<NOTE 2>: To minimize the number of memory leak false alarms (see | |
ac036724 | 2690 | L</PERL_DESTRUCT_LEVEL> for more information), you have to set the |
2691 | environment variable PERL_DESTRUCT_LEVEL to 2. | |
2692 | ||
2693 | For csh-like shells: | |
a958818a JH |
2694 | |
2695 | setenv PERL_DESTRUCT_LEVEL 2 | |
2696 | ||
ac036724 | 2697 | For Bourne-type shells: |
a958818a JH |
2698 | |
2699 | PERL_DESTRUCT_LEVEL=2 | |
2700 | export PERL_DESTRUCT_LEVEL | |
2701 | ||
ac036724 | 2702 | In Unixy environments you can also use the C<env> command: |
a958818a JH |
2703 | |
2704 | env PERL_DESTRUCT_LEVEL=2 valgrind ./perl -Ilib ... | |
a1b65709 | 2705 | |
37c0adeb JH |
2706 | B<NOTE 3>: There are known memory leaks when there are compile-time |
2707 | errors within eval or require, seeing C<S_doeval> in the call stack | |
2708 | is a good sign of these. Fixing these leaks is non-trivial, | |
2709 | unfortunately, but they must be fixed eventually. | |
2710 | ||
f50e5b73 MH |
2711 | B<NOTE 4>: L<DynaLoader> will not clean up after itself completely |
2712 | unless Perl is built with the Configure option | |
2713 | C<-Accflags=-DDL_UNLOAD_ALL_AT_EXIT>. | |
2714 | ||
902b9dbf MLF |
2715 | =head2 Rational Software's Purify |
2716 | ||
2717 | Purify is a commercial tool that is helpful in identifying | |
2718 | memory overruns, wild pointers, memory leaks and other such | |
2719 | badness. Perl must be compiled in a specific way for | |
2720 | optimal testing with Purify. Purify is available under | |
2721 | Windows NT, Solaris, HP-UX, SGI, and Siemens Unix. | |
2722 | ||
902b9dbf MLF |
2723 | =head2 Purify on Unix |
2724 | ||
2725 | On Unix, Purify creates a new Perl binary. To get the most | |
2726 | benefit out of Purify, you should create the perl to Purify | |
2727 | using: | |
2728 | ||
2729 | sh Configure -Accflags=-DPURIFY -Doptimize='-g' \ | |
2730 | -Uusemymalloc -Dusemultiplicity | |
2731 | ||
2732 | where these arguments mean: | |
2733 | ||
2734 | =over 4 | |
2735 | ||
2736 | =item -Accflags=-DPURIFY | |
2737 | ||
2738 | Disables Perl's arena memory allocation functions, as well as | |
2739 | forcing use of memory allocation functions derived from the | |
2740 | system malloc. | |
2741 | ||
2742 | =item -Doptimize='-g' | |
2743 | ||
2744 | Adds debugging information so that you see the exact source | |
2745 | statements where the problem occurs. Without this flag, all | |
2746 | you will see is the source filename of where the error occurred. | |
2747 | ||
2748 | =item -Uusemymalloc | |
2749 | ||
2750 | Disable Perl's malloc so that Purify can more closely monitor | |
2751 | allocations and leaks. Using Perl's malloc will make Purify | |
2752 | report most leaks in the "potential" leaks category. | |
2753 | ||
2754 | =item -Dusemultiplicity | |
2755 | ||
2756 | Enabling the multiplicity option allows perl to clean up | |
2757 | thoroughly when the interpreter shuts down, which reduces the | |
2758 | number of bogus leak reports from Purify. | |
2759 | ||
2760 | =back | |
2761 | ||
2762 | Once you've compiled a perl suitable for Purify'ing, then you | |
2763 | can just: | |
2764 | ||
07aa3531 | 2765 | make pureperl |
902b9dbf MLF |
2766 | |
2767 | which creates a binary named 'pureperl' that has been Purify'ed. | |
2768 | This binary is used in place of the standard 'perl' binary | |
2769 | when you want to debug Perl memory problems. | |
2770 | ||
2771 | As an example, to show any memory leaks produced during the | |
2772 | standard Perl testset you would create and run the Purify'ed | |
2773 | perl as: | |
2774 | ||
2775 | make pureperl | |
2776 | cd t | |
07aa3531 | 2777 | ../pureperl -I../lib harness |
902b9dbf MLF |
2778 | |
2779 | which would run Perl on test.pl and report any memory problems. | |
2780 | ||
2781 | Purify outputs messages in "Viewer" windows by default. If | |
2782 | you don't have a windowing environment or if you simply | |
2783 | want the Purify output to unobtrusively go to a log file | |
2784 | instead of to the interactive window, use these following | |
2785 | options to output to the log file "perl.log": | |
2786 | ||
2787 | setenv PURIFYOPTIONS "-chain-length=25 -windows=no \ | |
2788 | -log-file=perl.log -append-logfile=yes" | |
2789 | ||
2790 | If you plan to use the "Viewer" windows, then you only need this option: | |
2791 | ||
2792 | setenv PURIFYOPTIONS "-chain-length=25" | |
2793 | ||
c406981e JH |
2794 | In Bourne-type shells: |
2795 | ||
98631ff8 JL |
2796 | PURIFYOPTIONS="..." |
2797 | export PURIFYOPTIONS | |
c406981e JH |
2798 | |
2799 | or if you have the "env" utility: | |
2800 | ||
98631ff8 | 2801 | env PURIFYOPTIONS="..." ../pureperl ... |
c406981e | 2802 | |
902b9dbf MLF |
2803 | =head2 Purify on NT |
2804 | ||
2805 | Purify on Windows NT instruments the Perl binary 'perl.exe' | |
2806 | on the fly. There are several options in the makefile you | |
2807 | should change to get the most use out of Purify: | |
2808 | ||
2809 | =over 4 | |
2810 | ||
2811 | =item DEFINES | |
2812 | ||
2813 | You should add -DPURIFY to the DEFINES line so the DEFINES | |
2814 | line looks something like: | |
2815 | ||
07aa3531 | 2816 | DEFINES = -DWIN32 -D_CONSOLE -DNO_STRICT $(CRYPT_FLAG) -DPURIFY=1 |
902b9dbf MLF |
2817 | |
2818 | to disable Perl's arena memory allocation functions, as | |
2819 | well as to force use of memory allocation functions derived | |
2820 | from the system malloc. | |
2821 | ||
2822 | =item USE_MULTI = define | |
2823 | ||
2824 | Enabling the multiplicity option allows perl to clean up | |
2825 | thoroughly when the interpreter shuts down, which reduces the | |
2826 | number of bogus leak reports from Purify. | |
2827 | ||
2828 | =item #PERL_MALLOC = define | |
2829 | ||
2830 | Disable Perl's malloc so that Purify can more closely monitor | |
2831 | allocations and leaks. Using Perl's malloc will make Purify | |
2832 | report most leaks in the "potential" leaks category. | |
2833 | ||
2834 | =item CFG = Debug | |
2835 | ||
2836 | Adds debugging information so that you see the exact source | |
2837 | statements where the problem occurs. Without this flag, all | |
2838 | you will see is the source filename of where the error occurred. | |
2839 | ||
2840 | =back | |
2841 | ||
2842 | As an example, to show any memory leaks produced during the | |
2843 | standard Perl testset you would create and run Purify as: | |
2844 | ||
2845 | cd win32 | |
2846 | make | |
2847 | cd ../t | |
07aa3531 | 2848 | purify ../perl -I../lib harness |
902b9dbf MLF |
2849 | |
2850 | which would instrument Perl in memory, run Perl on test.pl, | |
2851 | then finally report any memory problems. | |
2852 | ||
7a834142 JH |
2853 | =head2 valgrind |
2854 | ||
2855 | The excellent valgrind tool can be used to find out both memory leaks | |
9df8f87f LB |
2856 | and illegal memory accesses. As of version 3.3.0, Valgrind only |
2857 | supports Linux on x86, x86-64 and PowerPC. The special "test.valgrind" | |
2858 | target can be used to run the tests under valgrind. Found errors | |
2859 | and memory leaks are logged in files named F<testfile.valgrind>. | |
07aa3531 JC |
2860 | |
2861 | Valgrind also provides a cachegrind tool, invoked on perl as: | |
2862 | ||
038c294a | 2863 | VG_OPTS=--tool=cachegrind make test.valgrind |
d44161bf MHM |
2864 | |
2865 | As system libraries (most notably glibc) are also triggering errors, | |
2866 | valgrind allows to suppress such errors using suppression files. The | |
2867 | default suppression file that comes with valgrind already catches a lot | |
2868 | of them. Some additional suppressions are defined in F<t/perl.supp>. | |
7a834142 JH |
2869 | |
2870 | To get valgrind and for more information see | |
2871 | ||
2872 | http://developer.kde.org/~sewardj/ | |
2873 | ||
f134cc4e | 2874 | =head2 Compaq's/Digital's/HP's Third Degree |
09187cb1 JH |
2875 | |
2876 | Third Degree is a tool for memory leak detection and memory access checks. | |
2877 | It is one of the many tools in the ATOM toolkit. The toolkit is only | |
2878 | available on Tru64 (formerly known as Digital UNIX formerly known as | |
2879 | DEC OSF/1). | |
2880 | ||
2881 | When building Perl, you must first run Configure with -Doptimize=-g | |
2882 | and -Uusemymalloc flags, after that you can use the make targets | |
51a35ef1 JH |
2883 | "perl.third" and "test.third". (What is required is that Perl must be |
2884 | compiled using the C<-g> flag, you may need to re-Configure.) | |
09187cb1 | 2885 | |
64cea5fd | 2886 | The short story is that with "atom" you can instrument the Perl |
83f0ef60 | 2887 | executable to create a new executable called F<perl.third>. When the |
4ae3d70a | 2888 | instrumented executable is run, it creates a log of dubious memory |
83f0ef60 | 2889 | traffic in file called F<perl.3log>. See the manual pages of atom and |
4ae3d70a JH |
2890 | third for more information. The most extensive Third Degree |
2891 | documentation is available in the Compaq "Tru64 UNIX Programmer's | |
2892 | Guide", chapter "Debugging Programs with Third Degree". | |
64cea5fd | 2893 | |
9c54ecba | 2894 | The "test.third" leaves a lot of files named F<foo_bar.3log> in the t/ |
64cea5fd JH |
2895 | subdirectory. There is a problem with these files: Third Degree is so |
2896 | effective that it finds problems also in the system libraries. | |
9c54ecba JH |
2897 | Therefore you should used the Porting/thirdclean script to cleanup |
2898 | the F<*.3log> files. | |
64cea5fd JH |
2899 | |
2900 | There are also leaks that for given certain definition of a leak, | |
2901 | aren't. See L</PERL_DESTRUCT_LEVEL> for more information. | |
2902 | ||
2903 | =head2 PERL_DESTRUCT_LEVEL | |
2904 | ||
a958818a JH |
2905 | If you want to run any of the tests yourself manually using e.g. |
2906 | valgrind, or the pureperl or perl.third executables, please note that | |
2907 | by default perl B<does not> explicitly cleanup all the memory it has | |
2908 | allocated (such as global memory arenas) but instead lets the exit() | |
2909 | of the whole program "take care" of such allocations, also known as | |
2910 | "global destruction of objects". | |
64cea5fd JH |
2911 | |
2912 | There is a way to tell perl to do complete cleanup: set the | |
2913 | environment variable PERL_DESTRUCT_LEVEL to a non-zero value. | |
2914 | The t/TEST wrapper does set this to 2, and this is what you | |
2915 | need to do too, if you don't want to see the "global leaks": | |
1f56d61a | 2916 | For example, for "third-degreed" Perl: |
64cea5fd | 2917 | |
1f56d61a | 2918 | env PERL_DESTRUCT_LEVEL=2 ./perl.third -Ilib t/foo/bar.t |
09187cb1 | 2919 | |
414f2397 RGS |
2920 | (Note: the mod_perl apache module uses also this environment variable |
2921 | for its own purposes and extended its semantics. Refer to the mod_perl | |
287a822c RGS |
2922 | documentation for more information. Also, spawned threads do the |
2923 | equivalent of setting this variable to the value 1.) | |
5a6c59ef DM |
2924 | |
2925 | If, at the end of a run you get the message I<N scalars leaked>, you can | |
fd0854ff DM |
2926 | recompile with C<-DDEBUG_LEAKING_SCALARS>, which will cause the addresses |
2927 | of all those leaked SVs to be dumped along with details as to where each | |
2928 | SV was originally allocated. This information is also displayed by | |
2929 | Devel::Peek. Note that the extra details recorded with each SV increases | |
2930 | memory usage, so it shouldn't be used in production environments. It also | |
2931 | converts C<new_SV()> from a macro into a real function, so you can use | |
2932 | your favourite debugger to discover where those pesky SVs were allocated. | |
414f2397 | 2933 | |
d7a2c63c MHM |
2934 | If you see that you're leaking memory at runtime, but neither valgrind |
2935 | nor C<-DDEBUG_LEAKING_SCALARS> will find anything, you're probably | |
2936 | leaking SVs that are still reachable and will be properly cleaned up | |
2937 | during destruction of the interpreter. In such cases, using the C<-Dm> | |
2938 | switch can point you to the source of the leak. If the executable was | |
2939 | built with C<-DDEBUG_LEAKING_SCALARS>, C<-Dm> will output SV allocations | |
2940 | in addition to memory allocations. Each SV allocation has a distinct | |
2941 | serial number that will be written on creation and destruction of the SV. | |
2942 | So if you're executing the leaking code in a loop, you need to look for | |
2943 | SVs that are created, but never destroyed between each cycle. If such an | |
2944 | SV is found, set a conditional breakpoint within C<new_SV()> and make it | |
2945 | break only when C<PL_sv_serial> is equal to the serial number of the | |
2946 | leaking SV. Then you will catch the interpreter in exactly the state | |
2947 | where the leaking SV is allocated, which is sufficient in many cases to | |
2948 | find the source of the leak. | |
2949 | ||
2950 | As C<-Dm> is using the PerlIO layer for output, it will by itself | |
2951 | allocate quite a bunch of SVs, which are hidden to avoid recursion. | |
2952 | You can bypass the PerlIO layer if you use the SV logging provided | |
2953 | by C<-DPERL_MEM_LOG> instead. | |
2954 | ||
46c6c7e2 JH |
2955 | =head2 PERL_MEM_LOG |
2956 | ||
10a879f5 JC |
2957 | If compiled with C<-DPERL_MEM_LOG>, both memory and SV allocations go |
2958 | through logging functions, which is handy for breakpoint setting. | |
2959 | ||
2960 | Unless C<-DPERL_MEM_LOG_NOIMPL> is also compiled, the logging | |
2e5b5004 | 2961 | functions read $ENV{PERL_MEM_LOG} to determine whether to log the |
10a879f5 JC |
2962 | event, and if so how: |
2963 | ||
2e5b5004 RGS |
2964 | $ENV{PERL_MEM_LOG} =~ /m/ Log all memory ops |
2965 | $ENV{PERL_MEM_LOG} =~ /s/ Log all SV ops | |
2966 | $ENV{PERL_MEM_LOG} =~ /t/ include timestamp in Log | |
2967 | $ENV{PERL_MEM_LOG} =~ /^(\d+)/ write to FD given (default is 2) | |
10a879f5 JC |
2968 | |
2969 | Memory logging is somewhat similar to C<-Dm> but is independent of | |
2970 | C<-DDEBUGGING>, and at a higher level; all uses of Newx(), Renew(), | |
2e5b5004 | 2971 | and Safefree() are logged with the caller's source code file and line |
10a879f5 JC |
2972 | number (and C function name, if supported by the C compiler). In |
2973 | contrast, C<-Dm> is directly at the point of C<malloc()>. SV logging | |
2974 | is similar. | |
2975 | ||
2976 | Since the logging doesn't use PerlIO, all SV allocations are logged | |
2977 | and no extra SV allocations are introduced by enabling the logging. | |
2978 | If compiled with C<-DDEBUG_LEAKING_SCALARS>, the serial number for | |
2979 | each SV allocation is also logged. | |
d7a2c63c | 2980 | |
51a35ef1 JH |
2981 | =head2 Profiling |
2982 | ||
2983 | Depending on your platform there are various of profiling Perl. | |
2984 | ||
2985 | There are two commonly used techniques of profiling executables: | |
10f58044 | 2986 | I<statistical time-sampling> and I<basic-block counting>. |
51a35ef1 JH |
2987 | |
2988 | The first method takes periodically samples of the CPU program | |
2989 | counter, and since the program counter can be correlated with the code | |
2990 | generated for functions, we get a statistical view of in which | |
2991 | functions the program is spending its time. The caveats are that very | |
2992 | small/fast functions have lower probability of showing up in the | |
2993 | profile, and that periodically interrupting the program (this is | |
2994 | usually done rather frequently, in the scale of milliseconds) imposes | |
2995 | an additional overhead that may skew the results. The first problem | |
2996 | can be alleviated by running the code for longer (in general this is a | |
2997 | good idea for profiling), the second problem is usually kept in guard | |
2998 | by the profiling tools themselves. | |
2999 | ||
10f58044 | 3000 | The second method divides up the generated code into I<basic blocks>. |
51a35ef1 JH |
3001 | Basic blocks are sections of code that are entered only in the |
3002 | beginning and exited only at the end. For example, a conditional jump | |
3003 | starts a basic block. Basic block profiling usually works by | |
10f58044 | 3004 | I<instrumenting> the code by adding I<enter basic block #nnnn> |
51a35ef1 JH |
3005 | book-keeping code to the generated code. During the execution of the |
3006 | code the basic block counters are then updated appropriately. The | |
3007 | caveat is that the added extra code can skew the results: again, the | |
3008 | profiling tools usually try to factor their own effects out of the | |
3009 | results. | |
3010 | ||
83f0ef60 JH |
3011 | =head2 Gprof Profiling |
3012 | ||
e1020413 | 3013 | gprof is a profiling tool available in many Unix platforms, |
51a35ef1 | 3014 | it uses F<statistical time-sampling>. |
83f0ef60 JH |
3015 | |
3016 | You can build a profiled version of perl called "perl.gprof" by | |
51a35ef1 JH |
3017 | invoking the make target "perl.gprof" (What is required is that Perl |
3018 | must be compiled using the C<-pg> flag, you may need to re-Configure). | |
3019 | Running the profiled version of Perl will create an output file called | |
3020 | F<gmon.out> is created which contains the profiling data collected | |
3021 | during the execution. | |
83f0ef60 JH |
3022 | |
3023 | The gprof tool can then display the collected data in various ways. | |
3024 | Usually gprof understands the following options: | |
3025 | ||
3026 | =over 4 | |
3027 | ||
3028 | =item -a | |
3029 | ||
3030 | Suppress statically defined functions from the profile. | |
3031 | ||
3032 | =item -b | |
3033 | ||
3034 | Suppress the verbose descriptions in the profile. | |
3035 | ||
3036 | =item -e routine | |
3037 | ||
3038 | Exclude the given routine and its descendants from the profile. | |
3039 | ||
3040 | =item -f routine | |
3041 | ||
3042 | Display only the given routine and its descendants in the profile. | |
3043 | ||
3044 | =item -s | |
3045 | ||
3046 | Generate a summary file called F<gmon.sum> which then may be given | |
3047 | to subsequent gprof runs to accumulate data over several runs. | |
3048 | ||
3049 | =item -z | |
3050 | ||
3051 | Display routines that have zero usage. | |
3052 | ||
3053 | =back | |
3054 | ||
3055 | For more detailed explanation of the available commands and output | |
3056 | formats, see your own local documentation of gprof. | |
3057 | ||
038c294a | 3058 | quick hint: |
07aa3531 | 3059 | |
289d61c2 JL |
3060 | $ sh Configure -des -Dusedevel -Doptimize='-pg' && make perl.gprof |
3061 | $ ./perl.gprof someprog # creates gmon.out in current directory | |
3062 | $ gprof ./perl.gprof > out | |
07aa3531 JC |
3063 | $ view out |
3064 | ||
51a35ef1 JH |
3065 | =head2 GCC gcov Profiling |
3066 | ||
10f58044 | 3067 | Starting from GCC 3.0 I<basic block profiling> is officially available |
51a35ef1 JH |
3068 | for the GNU CC. |
3069 | ||
3070 | You can build a profiled version of perl called F<perl.gcov> by | |
3071 | invoking the make target "perl.gcov" (what is required that Perl must | |
3072 | be compiled using gcc with the flags C<-fprofile-arcs | |
3073 | -ftest-coverage>, you may need to re-Configure). | |
3074 | ||
3075 | Running the profiled version of Perl will cause profile output to be | |
3076 | generated. For each source file an accompanying ".da" file will be | |
3077 | created. | |
3078 | ||
3079 | To display the results you use the "gcov" utility (which should | |
3080 | be installed if you have gcc 3.0 or newer installed). F<gcov> is | |
3081 | run on source code files, like this | |
3082 | ||
3083 | gcov sv.c | |
3084 | ||
3085 | which will cause F<sv.c.gcov> to be created. The F<.gcov> files | |
3086 | contain the source code annotated with relative frequencies of | |
3087 | execution indicated by "#" markers. | |
3088 | ||
3089 | Useful options of F<gcov> include C<-b> which will summarise the | |
3090 | basic block, branch, and function call coverage, and C<-c> which | |
3091 | instead of relative frequencies will use the actual counts. For | |
3092 | more information on the use of F<gcov> and basic block profiling | |
3093 | with gcc, see the latest GNU CC manual, as of GCC 3.0 see | |
3094 | ||
3095 | http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc.html | |
3096 | ||
3097 | and its section titled "8. gcov: a Test Coverage Program" | |
3098 | ||
3099 | http://gcc.gnu.org/onlinedocs/gcc-3.0/gcc_8.html#SEC132 | |
3100 | ||
07aa3531 JC |
3101 | quick hint: |
3102 | ||
3103 | $ sh Configure -des -Doptimize='-g' -Accflags='-fprofile-arcs -ftest-coverage' \ | |
3104 | -Aldflags='-fprofile-arcs -ftest-coverage' && make perl.gcov | |
3105 | $ rm -f regexec.c.gcov regexec.gcda | |
3106 | $ ./perl.gcov | |
3107 | $ gcov regexec.c | |
3108 | $ view regexec.c.gcov | |
3109 | ||
4ae3d70a JH |
3110 | =head2 Pixie Profiling |
3111 | ||
51a35ef1 JH |
3112 | Pixie is a profiling tool available on IRIX and Tru64 (aka Digital |
3113 | UNIX aka DEC OSF/1) platforms. Pixie does its profiling using | |
10f58044 | 3114 | I<basic-block counting>. |
4ae3d70a | 3115 | |
83f0ef60 | 3116 | You can build a profiled version of perl called F<perl.pixie> by |
51a35ef1 JH |
3117 | invoking the make target "perl.pixie" (what is required is that Perl |
3118 | must be compiled using the C<-g> flag, you may need to re-Configure). | |
3119 | ||
3120 | In Tru64 a file called F<perl.Addrs> will also be silently created, | |
3121 | this file contains the addresses of the basic blocks. Running the | |
3122 | profiled version of Perl will create a new file called "perl.Counts" | |
3123 | which contains the counts for the basic block for that particular | |
3124 | program execution. | |
4ae3d70a | 3125 | |
51a35ef1 | 3126 | To display the results you use the F<prof> utility. The exact |
4ae3d70a JH |
3127 | incantation depends on your operating system, "prof perl.Counts" in |
3128 | IRIX, and "prof -pixie -all -L. perl" in Tru64. | |
3129 | ||
6c41479b JH |
3130 | In IRIX the following prof options are available: |
3131 | ||
3132 | =over 4 | |
3133 | ||
3134 | =item -h | |
3135 | ||
3136 | Reports the most heavily used lines in descending order of use. | |
6e36760b | 3137 | Useful for finding the hotspot lines. |
6c41479b JH |
3138 | |
3139 | =item -l | |
3140 | ||
3141 | Groups lines by procedure, with procedures sorted in descending order of use. | |
3142 | Within a procedure, lines are listed in source order. | |
6e36760b | 3143 | Useful for finding the hotspots of procedures. |
6c41479b JH |
3144 | |
3145 | =back | |
3146 | ||
3147 | In Tru64 the following options are available: | |
3148 | ||
3149 | =over 4 | |
3150 | ||
3958b146 | 3151 | =item -p[rocedures] |
6c41479b | 3152 | |
3958b146 | 3153 | Procedures sorted in descending order by the number of cycles executed |
6e36760b | 3154 | in each procedure. Useful for finding the hotspot procedures. |
6c41479b JH |
3155 | (This is the default option.) |
3156 | ||
24000d2f | 3157 | =item -h[eavy] |
6c41479b | 3158 | |
6e36760b JH |
3159 | Lines sorted in descending order by the number of cycles executed in |
3160 | each line. Useful for finding the hotspot lines. | |
6c41479b | 3161 | |
24000d2f | 3162 | =item -i[nvocations] |
6c41479b | 3163 | |
6e36760b JH |
3164 | The called procedures are sorted in descending order by number of calls |
3165 | made to the procedures. Useful for finding the most used procedures. | |
6c41479b | 3166 | |
24000d2f | 3167 | =item -l[ines] |
6c41479b JH |
3168 | |
3169 | Grouped by procedure, sorted by cycles executed per procedure. | |
6e36760b | 3170 | Useful for finding the hotspots of procedures. |
6c41479b JH |
3171 | |
3172 | =item -testcoverage | |
3173 | ||
3174 | The compiler emitted code for these lines, but the code was unexecuted. | |
3175 | ||
24000d2f | 3176 | =item -z[ero] |
6c41479b JH |
3177 | |
3178 | Unexecuted procedures. | |
3179 | ||
aa500c9e | 3180 | =back |
6c41479b JH |
3181 | |
3182 | For further information, see your system's manual pages for pixie and prof. | |
4ae3d70a | 3183 | |
b8ddf6b3 SB |
3184 | =head2 Miscellaneous tricks |
3185 | ||
3186 | =over 4 | |
3187 | ||
3188 | =item * | |
3189 | ||
cc177e1a | 3190 | Those debugging perl with the DDD frontend over gdb may find the |
b8ddf6b3 SB |
3191 | following useful: |
3192 | ||
3193 | You can extend the data conversion shortcuts menu, so for example you | |
3194 | can display an SV's IV value with one click, without doing any typing. | |
3195 | To do that simply edit ~/.ddd/init file and add after: | |
3196 | ||
3197 | ! Display shortcuts. | |
3198 | Ddd*gdbDisplayShortcuts: \ | |
3199 | /t () // Convert to Bin\n\ | |
3200 | /d () // Convert to Dec\n\ | |
3201 | /x () // Convert to Hex\n\ | |
3202 | /o () // Convert to Oct(\n\ | |
3203 | ||
3204 | the following two lines: | |
3205 | ||
3206 | ((XPV*) (())->sv_any )->xpv_pv // 2pvx\n\ | |
3207 | ((XPVIV*) (())->sv_any )->xiv_iv // 2ivx | |
3208 | ||
3209 | so now you can do ivx and pvx lookups or you can plug there the | |
3210 | sv_peek "conversion": | |
3211 | ||
3212 | Perl_sv_peek(my_perl, (SV*)()) // sv_peek | |
3213 | ||
3214 | (The my_perl is for threaded builds.) | |
3215 | Just remember that every line, but the last one, should end with \n\ | |
3216 | ||
3217 | Alternatively edit the init file interactively via: | |
3218 | 3rd mouse button -> New Display -> Edit Menu | |
3219 | ||
3220 | Note: you can define up to 20 conversion shortcuts in the gdb | |
3221 | section. | |
3222 | ||
9965345d JH |
3223 | =item * |
3224 | ||
7e337ee0 JH |
3225 | If you see in a debugger a memory area mysteriously full of 0xABABABAB |
3226 | or 0xEFEFEFEF, you may be seeing the effect of the Poison() macros, | |
3227 | see L<perlclib>. | |
9965345d | 3228 | |
f1fac472 NC |
3229 | =item * |
3230 | ||
3231 | Under ithreads the optree is read only. If you want to enforce this, to check | |
3232 | for write accesses from buggy code, compile with C<-DPL_OP_SLAB_ALLOC> to | |
3233 | enable the OP slab allocator and C<-DPERL_DEBUG_READONLY_OPS> to enable code | |
3234 | that allocates op memory via C<mmap>, and sets it read-only at run time. | |
3235 | Any write access to an op results in a C<SIGBUS> and abort. | |
3236 | ||
3237 | This code is intended for development only, and may not be portable even to | |
3238 | all Unix variants. Also, it is an 80% solution, in that it isn't able to make | |
3239 | all ops read only. Specifically it | |
3240 | ||
3241 | =over | |
3242 | ||
3243 | =item 1 | |
3244 | ||
3245 | Only sets read-only on all slabs of ops at C<CHECK> time, hence ops allocated | |
3246 | later via C<require> or C<eval> will be re-write | |
3247 | ||
3248 | =item 2 | |
3249 | ||
3250 | Turns an entire slab of ops read-write if the refcount of any op in the slab | |
3251 | needs to be decreased. | |
3252 | ||
3253 | =item 3 | |
3254 | ||
3255 | Turns an entire slab of ops read-write if any op from the slab is freed. | |
3256 | ||
b8ddf6b3 SB |
3257 | =back |
3258 | ||
f1fac472 NC |
3259 | It's not possible to turn the slabs to read-only after an action requiring |
3260 | read-write access, as either can happen during op tree building time, so | |
3261 | there may still be legitimate write access. | |
3262 | ||
3263 | However, as an 80% solution it is still effective, as currently it catches | |
3264 | a write access during the generation of F<Config.pm>, which means that we | |
3265 | can't yet build F<perl> with this enabled. | |
3266 | ||
3267 | =back | |
3268 | ||
3269 | ||
955fec6b | 3270 | =head1 CONCLUSION |
a422fd2d | 3271 | |
955fec6b JH |
3272 | We've had a brief look around the Perl source, how to maintain quality |
3273 | of the source code, an overview of the stages F<perl> goes through | |
3274 | when it's running your code, how to use debuggers to poke at the Perl | |
3275 | guts, and finally how to analyse the execution of Perl. We took a very | |
3276 | simple problem and demonstrated how to solve it fully - with | |
3277 | documentation, regression tests, and finally a patch for submission to | |
3278 | p5p. Finally, we talked about how to use external tools to debug and | |
3279 | test Perl. | |
a422fd2d SC |
3280 | |
3281 | I'd now suggest you read over those references again, and then, as soon | |
3282 | as possible, get your hands dirty. The best way to learn is by doing, | |
07aa3531 | 3283 | so: |
a422fd2d SC |
3284 | |
3285 | =over 3 | |
3286 | ||
3287 | =item * | |
3288 | ||
3289 | Subscribe to perl5-porters, follow the patches and try and understand | |
3290 | them; don't be afraid to ask if there's a portion you're not clear on - | |
3291 | who knows, you may unearth a bug in the patch... | |
3292 | ||
3293 | =item * | |
3294 | ||
3295 | Keep up to date with the bleeding edge Perl distributions and get | |
3296 | familiar with the changes. Try and get an idea of what areas people are | |
3297 | working on and the changes they're making. | |
3298 | ||
3299 | =item * | |
3300 | ||
3e148164 | 3301 | Do read the README associated with your operating system, e.g. README.aix |
a1f349fd MB |
3302 | on the IBM AIX OS. Don't hesitate to supply patches to that README if |
3303 | you find anything missing or changed over a new OS release. | |
3304 | ||
3305 | =item * | |
3306 | ||
a422fd2d SC |
3307 | Find an area of Perl that seems interesting to you, and see if you can |
3308 | work out how it works. Scan through the source, and step over it in the | |
3309 | debugger. Play, poke, investigate, fiddle! You'll probably get to | |
3310 | understand not just your chosen area but a much wider range of F<perl>'s | |
3311 | activity as well, and probably sooner than you'd think. | |
3312 | ||
3313 | =back | |
3314 | ||
3315 | =over 3 | |
3316 | ||
3317 | =item I<The Road goes ever on and on, down from the door where it began.> | |
3318 | ||
3319 | =back | |
3320 | ||
64d9b66b | 3321 | If you can do these things, you've started on the long road to Perl porting. |
a422fd2d SC |
3322 | Thanks for wanting to help make Perl better - and happy hacking! |
3323 | ||
4ac71550 TC |
3324 | =head2 Metaphoric Quotations |
3325 | ||
3326 | If you recognized the quote about the Road above, you're in luck. | |
3327 | ||
3328 | Most software projects begin each file with a literal description of each | |
3329 | file's purpose. Perl instead begins each with a literary allusion to that | |
3330 | file's purpose. | |
3331 | ||
3332 | Like chapters in many books, all top-level Perl source files (along with a | |
3333 | few others here and there) begin with an epigramic inscription that alludes, | |
3334 | indirectly and metaphorically, to the material you're about to read. | |
3335 | ||
3336 | Quotations are taken from writings of J.R.R Tolkien pertaining to his | |
3337 | Legendarium, almost always from I<The Lord of the Rings>. Chapters and | |
3338 | page numbers are given using the following editions: | |
3339 | ||
3340 | =over 4 | |
3341 | ||
3342 | =item * | |
3343 | ||
3344 | I<The Hobbit>, by J.R.R. Tolkien. The hardcover, 70th-anniversary | |
3345 | edition of 2007 was used, published in the UK by Harper Collins Publishers | |
3346 | and in the US by the Houghton Mifflin Company. | |
3347 | ||
3348 | =item * | |
3349 | ||
3350 | I<The Lord of the Rings>, by J.R.R. Tolkien. The hardcover, | |
3351 | 50th-anniversary edition of 2004 was used, published in the UK by Harper | |
3352 | Collins Publishers and in the US by the Houghton Mifflin Company. | |
3353 | ||
3354 | =item * | |
3355 | ||
3356 | I<The Lays of Beleriand>, by J.R.R. Tolkien and published posthumously by his | |
3357 | son and literary executor, C.J.R. Tolkien, being the 3rd of the 12 volumes | |
3358 | in Christopher's mammoth I<History of Middle Earth>. Page numbers derive | |
3359 | from the hardcover edition, first published in 1983 by George Allen & | |
3360 | Unwin; no page numbers changed for the special 3-volume omnibus edition of | |
3361 | 2002 or the various trade-paper editions, all again now by Harper Collins | |
3362 | or Houghton Mifflin. | |
3363 | ||
3364 | =back | |
3365 | ||
3366 | Other JRRT books fair game for quotes would thus include I<The Adventures of | |
3367 | Tom Bombadil>, I<The Silmarillion>, I<Unfinished Tales>, and I<The Tale of | |
3368 | the Children of Hurin>, all but the first posthumously assembled by CJRT. | |
3369 | But I<The Lord of the Rings> itself is perfectly fine and probably best to | |
3370 | quote from, provided you can find a suitable quote there. | |
3371 | ||
3372 | So if you were to supply a new, complete, top-level source file to add to | |
3373 | Perl, you should conform to this peculiar practice by yourself selecting an | |
3374 | appropriate quotation from Tolkien, retaining the original spelling and | |
3375 | punctuation and using the same format the rest of the quotes are in. | |
3376 | Indirect and oblique is just fine; remember, it's a metaphor, so being meta | |
3377 | is, after all, what it's for. | |
3378 | ||
e8cd7eae GS |
3379 | =head1 AUTHOR |
3380 | ||
3381 | This document was written by Nathan Torkington, and is maintained by | |
3382 | the perl5-porters mailing list. | |
4ac71550 | 3383 | |
b16c2e4a RGS |
3384 | =head1 SEE ALSO |
3385 | ||
3386 | L<perlrepository> |