Commit | Line | Data |
---|---|---|
7711098a GS |
1 | =head1 NAME |
2 | ||
c3143508 | 3 | todo - Perl TO-DO list |
7711098a GS |
4 | |
5 | =head1 DESCRIPTION | |
e50bb9a1 | 6 | |
049aabcb | 7 | This is a list of wishes for Perl. The most up to date version of this file |
c3143508 | 8 | is at L<http://perl5.git.perl.org/perl.git/blob_plain/HEAD:/Porting/todo.pod> |
049aabcb NC |
9 | |
10 | The tasks we think are smaller or easier are listed first. Anyone is welcome | |
11 | to work on any of these, but it's a good idea to first contact | |
12 | I<perl5-porters@perl.org> to avoid duplication of effort, and to learn from | |
13 | any previous attempts. By all means contact a pumpking privately first if you | |
14 | prefer. | |
e50bb9a1 | 15 | |
0bdfc961 NC |
16 | Whilst patches to make the list shorter are most welcome, ideas to add to |
17 | the list are also encouraged. Check the perl5-porters archives for past | |
b4af8972 RB |
18 | ideas, and any discussion about them. One set of archives may be found at |
19 | L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/> | |
938c8732 | 20 | |
617eabfa NC |
21 | What can we offer you in return? Fame, fortune, and everlasting glory? Maybe |
22 | not, but if your patch is incorporated, then we'll add your name to the | |
23 | F<AUTHORS> file, which ships in the official distribution. How many other | |
24 | programming languages offer you 1 line of immortality? | |
938c8732 | 25 | |
0bdfc961 | 26 | =head1 Tasks that only need Perl knowledge |
e50bb9a1 | 27 | |
aa384da9 KW |
28 | =head2 Classify bug tickets by type |
29 | ||
1e5abc58 | 30 | Known bugs in Perl are tracked by L<https://rt.perl.org/> (which also |
aa384da9 | 31 | includes Perl 6). A summary can be found at |
1e5abc58 | 32 | L<https://rt.perl.org/NoAuth/perl5/Overview.html>. |
aa384da9 KW |
33 | It shows bugs classified by "type". However, the type of many of the |
34 | bugs is "unknown". This greatly lowers the chances of them getting | |
35 | fixed, as the number of open bugs is overwhelming -- too many to wade | |
36 | through for someone to try to find the bugs in the parts of | |
37 | Perl that s/he knows well enough to try to fix. This task involves | |
38 | going through these bugs and classifying them into one or more types. | |
39 | ||
40 | =head2 Ongoing: investigate new bug reports | |
41 | ||
42 | When a bug report is filed, it would be very helpful to have someone do | |
43 | a quick investigation to see if it is a real problem, and to reply to | |
44 | the poster about it, asking for example code that reproduces the | |
45 | problem. Such code should be added to the test suite as TODO tests, and | |
46 | the ticket should be classified by type. To get started on this task, | |
47 | look at the tickets that are marked as "New Issues" in | |
1e5abc58 | 48 | L<https://rt.perl.org/NoAuth/perl5/Overview.html>. |
aa384da9 | 49 | |
de2b17d8 NC |
50 | =head2 Migrate t/ from custom TAP generation |
51 | ||
52 | Many tests below F<t/> still generate TAP by "hand", rather than using library | |
96090e4f | 53 | functions. As explained in L<perlhack/TESTING>, tests in F<t/> are |
de2b17d8 NC |
54 | written in a particular way to test that more complex constructions actually |
55 | work before using them routinely. Hence they don't use C<Test::More>, but | |
56 | instead there is an intentionally simpler library, F<t/test.pl>. However, | |
57 | quite a few tests in F<t/> have not been refactored to use it. Refactoring | |
58 | any of these tests, one at a time, is a useful thing TODO. | |
59 | ||
f3227d3b JK |
60 | The subdirectories F<base>, F<cmd>, F<comp> and F<opbasic>, that contain the |
61 | most basic tests, should be excluded from this task. | |
0d8e5a42 | 62 | |
0be987a2 NC |
63 | =head2 Automate perldelta generation |
64 | ||
65 | The perldelta file accompanying each release summaries the major changes. | |
66 | It's mostly manually generated currently, but some of that could be | |
67 | automated with a bit of perl, specifically the generation of | |
68 | ||
69 | =over | |
70 | ||
71 | =item Modules and Pragmata | |
72 | ||
73 | =item New Documentation | |
74 | ||
75 | =item New Tests | |
76 | ||
77 | =back | |
78 | ||
79 | See F<Porting/how_to_write_a_perldelta.pod> for details. | |
80 | ||
0bdfc961 | 81 | =head2 Make Schwern poorer |
e50bb9a1 | 82 | |
613bd4f7 | 83 | We should have tests for everything. When all the core's modules are tested, |
0bdfc961 NC |
84 | Schwern has promised to donate to $500 to TPF. We may need volunteers to |
85 | hold him upside down and shake vigorously in order to actually extract the | |
86 | cash. | |
3958b146 | 87 | |
831efc5a JK |
88 | =head2 Write descriptions for all tests |
89 | ||
90 | Many individual tests in the test suite lack descriptions (or names, or labels | |
91 | -- call them what you will). Many files completely lack descriptions, meaning | |
92 | that the only output you get is the test numbers. If all tests had | |
93 | descriptions, understanding what the tests are testing and why they sometimes | |
94 | fail would both get a whole lot easier. | |
95 | ||
0bdfc961 | 96 | =head2 Improve the coverage of the core tests |
e50bb9a1 | 97 | |
e1020413 | 98 | Use Devel::Cover to ascertain the core modules' test coverage, then add |
02f21748 | 99 | tests that are currently missing. |
30222c0f | 100 | |
0bdfc961 | 101 | =head2 test B |
e50bb9a1 | 102 | |
0bdfc961 | 103 | A full test suite for the B module would be nice. |
e50bb9a1 | 104 | |
0bdfc961 | 105 | =head2 A decent benchmark |
e50bb9a1 | 106 | |
617eabfa | 107 | C<perlbench> seems impervious to any recent changes made to the perl core. It |
0bdfc961 NC |
108 | would be useful to have a reasonable general benchmarking suite that roughly |
109 | represented what current perl programs do, and measurably reported whether | |
110 | tweaks to the core improve, degrade or don't really affect performance, to | |
111 | guide people attempting to optimise the guts of perl. Gisle would welcome | |
4e1c9055 NC |
112 | new tests for perlbench. Steffen Schwingon would welcome help with |
113 | L<Benchmark::Perl::Formance> | |
6168cf99 | 114 | |
0bdfc961 | 115 | =head2 fix tainting bugs |
6168cf99 | 116 | |
82046965 FC |
117 | Fix the bugs revealed by running the test suite with the C<-t> switch. |
118 | Setting the TEST_ARGS environment variable to C<-taintwarn> will accomplish | |
119 | this. | |
e50bb9a1 | 120 | |
0bdfc961 | 121 | =head2 Dual life everything |
e50bb9a1 | 122 | |
0bdfc961 NC |
123 | As part of the "dists" plan, anything that doesn't belong in the smallest perl |
124 | distribution needs to be dual lifed. Anything else can be too. Figure out what | |
125 | changes would be needed to package that module and its tests up for CPAN, and | |
126 | do so. Test it with older perl releases, and fix the problems you find. | |
e50bb9a1 | 127 | |
a393eb28 RGS |
128 | To make a minimal perl distribution, it's useful to look at |
129 | F<t/lib/commonsense.t>. | |
130 | ||
0bdfc961 | 131 | =head2 POSIX memory footprint |
e50bb9a1 | 132 | |
0bdfc961 NC |
133 | Ilya observed that use POSIX; eats memory like there's no tomorrow, and at |
134 | various times worked to cut it down. There is probably still fat to cut out - | |
135 | for example POSIX passes Exporter some very memory hungry data structures. | |
e50bb9a1 | 136 | |
8c422da5 NC |
137 | =head2 makedef.pl and conditional compilation |
138 | ||
139 | The script F<makedef.pl> that generates the list of exported symbols on | |
140 | platforms which need this. Functions are declared in F<embed.fnc>, variables | |
141 | in F<intrpvar.h>. Quite a few of the functions and variables are conditionally | |
142 | declared there, using C<#ifdef>. However, F<makedef.pl> doesn't understand the | |
143 | C macros, so the rules about which symbols are present when is duplicated in | |
144 | the Perl code. Writing things twice is bad, m'kay. It would be good to teach | |
145 | F<.pl> to understand the conditional compilation, and hence remove the | |
146 | duplication, and the mistakes it has caused. | |
e50bb9a1 | 147 | |
801de10e NC |
148 | =head2 use strict; and AutoLoad |
149 | ||
150 | Currently if you write | |
151 | ||
152 | package Whack; | |
153 | use AutoLoader 'AUTOLOAD'; | |
154 | use strict; | |
155 | 1; | |
156 | __END__ | |
157 | sub bloop { | |
158 | print join (' ', No, strict, here), "!\n"; | |
159 | } | |
160 | ||
161 | then C<use strict;> isn't in force within the autoloaded subroutines. It would | |
162 | be more consistent (and less surprising) to arrange for all lexical pragmas | |
163 | in force at the __END__ block to be in force within each autoloaded subroutine. | |
164 | ||
773b3597 RGS |
165 | There's a similar problem with SelfLoader. |
166 | ||
91d0cbf6 NC |
167 | =head2 profile installman |
168 | ||
169 | The F<installman> script is slow. All it is doing text processing, which we're | |
170 | told is something Perl is good at. So it would be nice to know what it is doing | |
171 | that is taking so much CPU, and where possible address it. | |
172 | ||
c69ca1d4 | 173 | =head2 enable lexical enabling/disabling of individual warnings |
a9ed9b74 JV |
174 | |
175 | Currently, warnings can only be enabled or disabled by category. There | |
176 | are times when it would be useful to quash a single warning, not a | |
177 | whole category. | |
91d0cbf6 | 178 | |
85234543 KW |
179 | =head2 document diagnostics |
180 | ||
181 | Many diagnostic messages are not currently documented. The list is at the end | |
182 | of t/porting/diag.t. | |
183 | ||
aa384da9 KW |
184 | =head2 Write TODO tests for open bugs |
185 | ||
186 | Sometimes bugs get fixed as a side effect of something else, and | |
187 | the bug remains open because no one realizes that it has been fixed. | |
188 | Ideally, every open bug should have a TODO test in the core test suite. | |
189 | ||
98b8e361 FC |
190 | =head2 deparse warnings nicely |
191 | ||
192 | Currently Deparse punts on deparsing the bitmask for warnings, which it | |
193 | dumps uglily as-is. Try running this: | |
194 | ||
195 | $ ./perl -Ilib -MO=Deparse -e 'use warnings "pipe"; die' | |
196 | ||
197 | Deparse.pm could use the package variables in warnings.pm that warnings.pm | |
198 | itself uses to convert the list passed to it into a bitfield. Deparse just | |
199 | needs to reverse that. | |
200 | ||
201 | =head2 test and fix Deparse with perl's test suite | |
202 | ||
203 | If you run perl's tests with the TEST_ARGS environment variable set to | |
204 | C<-deparse> (e.g., run C<TEST=-deparse make test>), each test file will be | |
205 | deparsed and the deparsed output will be run. Currently there are many | |
206 | failures, which ought to be fixed. There is in F<Porting/deparse-skips.txt> | |
207 | a list of tests known to fail, but it is out of date. Updating it would | |
208 | also help. | |
209 | ||
210 | This is an incremental task. Every small bit helps. It is also a task that | |
211 | may never end. As new tests are added, they tickle corner cases that | |
212 | B::Deparse cannot yet handle correctly. | |
213 | ||
214 | This task I<may> need a bit of perl guts knowledge. But what changes need | |
215 | to be made is usually easy to see by dumping op trees with B::Concise: | |
216 | ||
217 | $ ./perl -Ilib -MO=Concise -e 'foo(); print @_; die $$_' | |
218 | ||
219 | and adjusting B::Deparse to handle whatever you see B::Concise produce. | |
220 | This is also a good way to I<learn> how perl's op trees work. | |
221 | ||
0bdfc961 | 222 | =head1 Tasks that need a little sysadmin-type knowledge |
e50bb9a1 | 223 | |
0bdfc961 NC |
224 | Or if you prefer, tasks that you would learn from, and broaden your skills |
225 | base... | |
e50bb9a1 | 226 | |
cd793d32 | 227 | =head2 make HTML install work |
e50bb9a1 | 228 | |
78b489b0 | 229 | There is an C<install.html> target in the Makefile. It's marked as |
adebf063 NC |
230 | "experimental". It would be good to get this tested, make it work reliably, and |
231 | remove the "experimental" tag. This would include | |
232 | ||
233 | =over 4 | |
234 | ||
235 | =item 1 | |
236 | ||
237 | Checking that cross linking between various parts of the documentation works. | |
238 | In particular that links work between the modules (files with POD in F<lib/>) | |
239 | and the core documentation (files in F<pod/>) | |
240 | ||
241 | =item 2 | |
242 | ||
78b489b0 NC |
243 | Improving the code that split C<perlfunc> into chunks, preferably with |
244 | general case code added to L<Pod::Functions> that could be used elsewhere. | |
245 | ||
617eabfa NC |
246 | Challenges here are correctly identifying the groups of functions that go |
247 | together, and making the right named external cross-links point to the right | |
78b489b0 NC |
248 | page. Currently this works reasonably well in the general case, and correctly |
249 | parses two or more C<=items> giving the different parameter lists for the | |
250 | same function, such used by C<substr>. However it fails completely where | |
251 | I<different> functions are listed as a sequence of C<=items> but share the | |
252 | same description. All the functions from C<getpwnam> to C<endprotoent> have | |
253 | individual stub pages, with only the page for C<endservent> holding the | |
254 | description common to all. Likewise C<q>, C<qq> and C<qw> have stub pages, | |
255 | instead of sharing the body of C<qx>. | |
256 | ||
257 | Note also the current code isn't ideal with the two forms of C<select>, mushing | |
258 | them both into one F<select.html> with the two descriptions run together. | |
259 | Fixing this may well be a special case. | |
adebf063 NC |
260 | |
261 | =back | |
3a89a73c | 262 | |
0bdfc961 NC |
263 | =head2 compressed man pages |
264 | ||
265 | Be able to install them. This would probably need a configure test to see how | |
266 | the system does compressed man pages (same directory/different directory? | |
267 | same filename/different filename), as well as tweaking the F<installman> script | |
268 | to compress as necessary. | |
269 | ||
30222c0f NC |
270 | =head2 Add a code coverage target to the Makefile |
271 | ||
272 | Make it easy for anyone to run Devel::Cover on the core's tests. The steps | |
273 | to do this manually are roughly | |
274 | ||
275 | =over 4 | |
276 | ||
277 | =item * | |
278 | ||
279 | do a normal C<Configure>, but include Devel::Cover as a module to install | |
f11a3063 | 280 | (see L<INSTALL> for how to do this) |
30222c0f NC |
281 | |
282 | =item * | |
283 | ||
284 | make perl | |
285 | ||
286 | =item * | |
287 | ||
f185f654 | 288 | cd t; HARNESS_PERL_SWITCHES=-MDevel::Cover ./perl -I../lib harness |
30222c0f NC |
289 | |
290 | =item * | |
291 | ||
292 | Process the resulting Devel::Cover database | |
293 | ||
294 | =back | |
295 | ||
296 | This just give you the coverage of the F<.pm>s. To also get the C level | |
297 | coverage you need to | |
298 | ||
299 | =over 4 | |
300 | ||
301 | =item * | |
302 | ||
303 | Additionally tell C<Configure> to use the appropriate C compiler flags for | |
304 | C<gcov> | |
305 | ||
306 | =item * | |
307 | ||
308 | make perl.gcov | |
309 | ||
310 | (instead of C<make perl>) | |
311 | ||
312 | =item * | |
313 | ||
314 | After running the tests run C<gcov> to generate all the F<.gcov> files. | |
315 | (Including down in the subdirectories of F<ext/> | |
316 | ||
317 | =item * | |
318 | ||
319 | (From the top level perl directory) run C<gcov2perl> on all the C<.gcov> files | |
320 | to get their stats into the cover_db directory. | |
321 | ||
322 | =item * | |
323 | ||
324 | Then process the Devel::Cover database | |
325 | ||
326 | =back | |
327 | ||
328 | It would be good to add a single switch to C<Configure> to specify that you | |
329 | wanted to perform perl level coverage, and another to specify C level | |
330 | coverage, and have C<Configure> and the F<Makefile> do all the right things | |
331 | automatically. | |
332 | ||
02f21748 | 333 | =head2 Make Config.pm cope with differences between built and installed perl |
0bdfc961 NC |
334 | |
335 | Quite often vendors ship a perl binary compiled with their (pay-for) | |
336 | compilers. People install a free compiler, such as gcc. To work out how to | |
337 | build extensions, Perl interrogates C<%Config>, so in this situation | |
338 | C<%Config> describes compilers that aren't there, and extension building | |
339 | fails. This forces people into choosing between re-compiling perl themselves | |
340 | using the compiler they have, or only using modules that the vendor ships. | |
341 | ||
342 | It would be good to find a way teach C<Config.pm> about the installation setup, | |
343 | possibly involving probing at install time or later, so that the C<%Config> in | |
344 | a binary distribution better describes the installed machine, when the | |
345 | installed machine differs from the build machine in some significant way. | |
346 | ||
728f4ecd NC |
347 | =head2 linker specification files |
348 | ||
349 | Some platforms mandate that you provide a list of a shared library's external | |
350 | symbols to the linker, so the core already has the infrastructure in place to | |
4e1c9055 NC |
351 | do this for generating shared perl libraries. Florian Ragwitz has been working |
352 | to offer this for the GNU toolchain, to allow Unix users to test that the | |
728f4ecd | 353 | export list is correct, and to build a perl that does not pollute the global |
32d539f5 | 354 | namespace with private symbols, and will fail in the same way as msvc or mingw |
4e1c9055 | 355 | builds or when using PERL_DL_NONLAZY=1. See the branch smoke-me/rafl/ld_export |
728f4ecd | 356 | |
a229ae3b RGS |
357 | =head2 Cross-compile support |
358 | ||
4e1c9055 NC |
359 | We get requests for "how to cross compile Perl". The vast majority of these |
360 | seem to be for a couple of scenarios: | |
361 | ||
362 | =over 4 | |
363 | ||
364 | =item * | |
365 | ||
366 | Platforms that could build natively using F<./Configure> (I<e.g.> Linux or | |
367 | NetBSD on MIPS or ARM) but people want to use a beefier machine (and on the | |
368 | same OS) to build more easily. | |
369 | ||
370 | =item * | |
371 | ||
372 | Platforms that can't build natively, but no (significant) porting changes | |
373 | are needed to our current source code. Prime example of this is Android. | |
374 | ||
375 | =back | |
376 | ||
377 | There are several scripts and tools for cross-compiling perl for other | |
378 | platforms. However, these are somewhat inconsistent and scattered across the | |
379 | codebase, none are documented well, none are clearly flexible enough to | |
c5fb089a | 380 | be confident that they can support any TARGET/HOST platform pair other than |
4e1c9055 NC |
381 | that which they were developed on, and it's not clear how bitrotted they are. |
382 | ||
383 | For example, C<Configure> understands C<-Dusecrosscompile> option. This option | |
a229ae3b | 384 | arranges for building C<miniperl> for TARGET machine, so this C<miniperl> is |
4e1c9055 NC |
385 | assumed then to be copied to TARGET machine and used as a replacement of |
386 | full C<perl> executable. This code is almost 10 years old. Meanwhile, the | |
387 | F<Cross/> directory contains two different approaches for cross compiling to | |
388 | ARM Linux targets, relying on hand curated F<config.sh> files, but that code | |
389 | is getting on for 5 years old, and requires insider knowledge of perl's | |
390 | build system to draft a F<config.sh> for a new platform. | |
391 | ||
c5fb089a | 392 | Jess Robinson has submitted a grant to TPF to work on cleaning this up. |
0bdfc961 | 393 | |
98fca0e8 NC |
394 | =head2 Split "linker" from "compiler" |
395 | ||
396 | Right now, Configure probes for two commands, and sets two variables: | |
397 | ||
398 | =over 4 | |
399 | ||
b91dd380 | 400 | =item * C<cc> (in F<cc.U>) |
98fca0e8 NC |
401 | |
402 | This variable holds the name of a command to execute a C compiler which | |
403 | can resolve multiple global references that happen to have the same | |
404 | name. Usual values are F<cc> and F<gcc>. | |
405 | Fervent ANSI compilers may be called F<c89>. AIX has F<xlc>. | |
406 | ||
b91dd380 | 407 | =item * C<ld> (in F<dlsrc.U>) |
98fca0e8 NC |
408 | |
409 | This variable indicates the program to be used to link | |
410 | libraries for dynamic loading. On some systems, it is F<ld>. | |
411 | On ELF systems, it should be C<$cc>. Mostly, we'll try to respect | |
412 | the hint file setting. | |
413 | ||
414 | =back | |
415 | ||
8d159ec1 NC |
416 | There is an implicit historical assumption from around Perl5.000alpha |
417 | something, that C<$cc> is also the correct command for linking object files | |
418 | together to make an executable. This may be true on Unix, but it's not true | |
419 | on other platforms, and there are a maze of work arounds in other places (such | |
420 | as F<Makefile.SH>) to cope with this. | |
98fca0e8 NC |
421 | |
422 | Ideally, we should create a new variable to hold the name of the executable | |
423 | linker program, probe for it in F<Configure>, and centralise all the special | |
424 | case logic there or in hints files. | |
425 | ||
426 | A small bikeshed issue remains - what to call it, given that C<$ld> is already | |
8d159ec1 NC |
427 | taken (arguably for the wrong thing now, but on SunOS 4.1 it is the command |
428 | for creating dynamically-loadable modules) and C<$link> could be confused with | |
429 | the Unix command line executable of the same name, which does something | |
430 | completely different. Andy Dougherty makes the counter argument "In parrot, I | |
431 | tried to call the command used to link object files and libraries into an | |
432 | executable F<link>, since that's what my vaguely-remembered DOS and VMS | |
433 | experience suggested. I don't think any real confusion has ensued, so it's | |
434 | probably a reasonable name for perl5 to use." | |
98fca0e8 NC |
435 | |
436 | "Alas, I've always worried that introducing it would make things worse, | |
437 | since now the module building utilities would have to look for | |
438 | C<$Config{link}> and institute a fall-back plan if it weren't found." | |
8d159ec1 NC |
439 | Although I can see that as confusing, given that C<$Config{d_link}> is true |
440 | when (hard) links are available. | |
98fca0e8 | 441 | |
75585ce3 SP |
442 | =head2 Configure Windows using PowerShell |
443 | ||
444 | Currently, Windows uses hard-coded config files based to build the | |
445 | config.h for compiling Perl. Makefiles are also hard-coded and need to be | |
446 | hand edited prior to building Perl. While this makes it easy to create a perl.exe | |
447 | that works across multiple Windows versions, being able to accurately | |
448 | configure a perl.exe for a specific Windows versions and VS C++ would be | |
449 | a nice enhancement. With PowerShell available on Windows XP and up, this | |
450 | may now be possible. Step 1 might be to investigate whether this is possible | |
451 | and use this to clean up our current makefile situation. Step 2 would be to | |
452 | see if there would be a way to use our existing metaconfig units to configure a | |
453 | Windows Perl or whether we go in a separate direction and make it so. Of | |
454 | course, we all know what step 3 is. | |
455 | ||
0bdfc961 NC |
456 | =head1 Tasks that need a little C knowledge |
457 | ||
458 | These tasks would need a little C knowledge, but don't need any specific | |
459 | background or experience with XS, or how the Perl interpreter works | |
460 | ||
3d826b29 NC |
461 | =head2 Weed out needless PERL_UNUSED_ARG |
462 | ||
463 | The C code uses the macro C<PERL_UNUSED_ARG> to stop compilers warning about | |
464 | unused arguments. Often the arguments can't be removed, as there is an | |
465 | external constraint that determines the prototype of the function, so this | |
466 | approach is valid. However, there are some cases where C<PERL_UNUSED_ARG> | |
467 | could be removed. Specifically | |
468 | ||
469 | =over 4 | |
470 | ||
471 | =item * | |
472 | ||
473 | The prototypes of (nearly all) static functions can be changed | |
474 | ||
475 | =item * | |
476 | ||
477 | Unused arguments generated by short cut macros are wasteful - the short cut | |
478 | macro used can be changed. | |
479 | ||
480 | =back | |
481 | ||
bcbaa2d5 RGS |
482 | =head2 -Duse32bit* |
483 | ||
484 | Natively 64-bit systems need neither -Duse64bitint nor -Duse64bitall. | |
485 | On these systems, it might be the default compilation mode, and there | |
486 | is currently no guarantee that passing no use64bitall option to the | |
487 | Configure process will build a 32bit perl. Implementing -Duse32bit* | |
54a814f4 | 488 | options would be nice for perl 5.26.0. |
bcbaa2d5 | 489 | |
fee0a0f7 | 490 | =head2 Profile Perl - am I hot or not? |
62403a3c | 491 | |
fee0a0f7 NC |
492 | The Perl source code is stable enough that it makes sense to profile it, |
493 | identify and optimise the hotspots. It would be good to measure the | |
494 | performance of the Perl interpreter using free tools such as cachegrind, | |
495 | gprof, and dtrace, and work to reduce the bottlenecks they reveal. | |
496 | ||
497 | As part of this, the idea of F<pp_hot.c> is that it contains the I<hot> ops, | |
498 | the ops that are most commonly used. The idea is that by grouping them, their | |
499 | object code will be adjacent in the executable, so they have a greater chance | |
500 | of already being in the CPU cache (or swapped in) due to being near another op | |
501 | already in use. | |
62403a3c NC |
502 | |
503 | Except that it's not clear if these really are the most commonly used ops. So | |
fee0a0f7 NC |
504 | as part of exercising your skills with coverage and profiling tools you might |
505 | want to determine what ops I<really> are the most commonly used. And in turn | |
506 | suggest evictions and promotions to achieve a better F<pp_hot.c>. | |
62403a3c | 507 | |
91d0cbf6 NC |
508 | One piece of Perl code that might make a good testbed is F<installman>. |
509 | ||
a229ae3b | 510 | =head2 Improve win32/wince.c |
0bdfc961 | 511 | |
a229ae3b | 512 | Currently, numerous functions look virtually, if not completely, |
c23989d1 | 513 | identical in both F<win32/wince.c> and F<win32/win32.c> files, which can't |
6d71adcd NC |
514 | be good. |
515 | ||
c5b31784 SH |
516 | =head2 Use secure CRT functions when building with VC8 on Win32 |
517 | ||
518 | Visual C++ 2005 (VC++ 8.x) deprecated a number of CRT functions on the basis | |
519 | that they were "unsafe" and introduced differently named secure versions of | |
520 | them as replacements, e.g. instead of writing | |
521 | ||
522 | FILE* f = fopen(__FILE__, "r"); | |
523 | ||
524 | one should now write | |
525 | ||
526 | FILE* f; | |
527 | errno_t err = fopen_s(&f, __FILE__, "r"); | |
528 | ||
529 | Currently, the warnings about these deprecations have been disabled by adding | |
530 | -D_CRT_SECURE_NO_DEPRECATE to the CFLAGS. It would be nice to remove that | |
531 | warning suppressant and actually make use of the new secure CRT functions. | |
532 | ||
533 | There is also a similar issue with POSIX CRT function names like fileno having | |
534 | been deprecated in favour of ISO C++ conformant names like _fileno. These | |
26a6faa8 | 535 | warnings are also currently suppressed by adding -D_CRT_NONSTDC_NO_DEPRECATE. It |
c5b31784 SH |
536 | might be nice to do as Microsoft suggest here too, although, unlike the secure |
537 | functions issue, there is presumably little or no benefit in this case. | |
538 | ||
038ae9a4 SH |
539 | =head2 Fix POSIX::access() and chdir() on Win32 |
540 | ||
541 | These functions currently take no account of DACLs and therefore do not behave | |
542 | correctly in situations where access is restricted by DACLs (as opposed to the | |
543 | read-only attribute). | |
544 | ||
545 | Furthermore, POSIX::access() behaves differently for directories having the | |
546 | read-only attribute set depending on what CRT library is being used. For | |
547 | example, the _access() function in the VC6 and VC7 CRTs (wrongly) claim that | |
548 | such directories are not writable, whereas in fact all directories are writable | |
549 | unless access is denied by DACLs. (In the case of directories, the read-only | |
550 | attribute actually only means that the directory cannot be deleted.) This CRT | |
551 | bug is fixed in the VC8 and VC9 CRTs (but, of course, the directory may still | |
552 | not actually be writable if access is indeed denied by DACLs). | |
553 | ||
554 | For the chdir() issue, see ActiveState bug #74552: | |
b4af8972 | 555 | L<http://bugs.activestate.com/show_bug.cgi?id=74552> |
038ae9a4 SH |
556 | |
557 | Therefore, DACLs should be checked both for consistency across CRTs and for | |
558 | the correct answer. | |
559 | ||
560 | (Note that perl's -w operator should not be modified to check DACLs. It has | |
561 | been written so that it reflects the state of the read-only attribute, even | |
562 | for directories (whatever CRT is being used), for symmetry with chmod().) | |
563 | ||
8964cfe0 NC |
564 | =head2 Arenas for GPs? For MAGIC? |
565 | ||
566 | C<struct gp> and C<struct magic> are both currently allocated by C<malloc>. | |
567 | It might be a speed or memory saving to change to using arenas. Or it might | |
568 | not. It would need some suitable benchmarking first. In particular, C<GP>s | |
569 | can probably be changed with minimal compatibility impact (probably nothing | |
570 | outside of the core, or even outside of F<gv.c> allocates them), but they | |
571 | probably aren't allocated/deallocated often enough for a speed saving. Whereas | |
572 | C<MAGIC> is allocated/deallocated more often, but in turn, is also something | |
573 | more externally visible, so changing the rules here may bite external code. | |
574 | ||
3880c8ec NC |
575 | =head2 Shared arenas |
576 | ||
577 | Several SV body structs are now the same size, notably PVMG and PVGV, PVAV and | |
578 | PVHV, and PVCV and PVFM. It should be possible to allocate and return same | |
579 | sized bodies from the same actual arena, rather than maintaining one arena for | |
580 | each. This could save 4-6K per thread, of memory no longer tied up in the | |
581 | not-yet-allocated part of an arena. | |
582 | ||
8964cfe0 | 583 | |
6d71adcd NC |
584 | =head1 Tasks that need a knowledge of XS |
585 | ||
586 | These tasks would need C knowledge, and roughly the level of knowledge of | |
587 | the perl API that comes from writing modules that use XS to interface to | |
588 | C. | |
589 | ||
e851c105 DG |
590 | =head2 Write an XS cookbook |
591 | ||
592 | Create pod/perlxscookbook.pod with short, task-focused 'recipes' in XS that | |
593 | demonstrate common tasks and good practices. (Some of these might be | |
594 | extracted from perlguts.) The target audience should be XS novices, who need | |
595 | more examples than perlguts but something less overwhelming than perlapi. | |
596 | Recipes should provide "one pretty good way to do it" instead of TIMTOWTDI. | |
597 | ||
5b7d14ff DG |
598 | Rather than focusing on interfacing Perl to C libraries, such a cookbook |
599 | should probably focus on how to optimize Perl routines by re-writing them | |
600 | in XS. This will likely be more motivating to those who mostly work in | |
601 | Perl but are looking to take the next step into XS. | |
602 | ||
603 | Deconstructing and explaining some simpler XS modules could be one way to | |
604 | bootstrap a cookbook. (List::Util? Class::XSAccessor? Tree::Ternary_XS?) | |
605 | Another option could be deconstructing the implementation of some simpler | |
606 | functions in op.c. | |
607 | ||
0b162fb0 | 608 | =head2 Document how XSUBs can use C<cv_set_call_checker> to inline themselves as OPs |
05fb4e20 NC |
609 | |
610 | For a simple XSUB, often the subroutine dispatch takes more time than the | |
0b162fb0 NC |
611 | XSUB itself. v5.14.0 now allows XSUBs to register a function which will be |
612 | called when the parser is finished building an C<entersub> op which calls | |
613 | them. | |
614 | ||
615 | Registration is done with C<Perl_cv_set_call_checker>, is documented at the | |
616 | API level in L<perlapi>, and L<perl5140delta/Custom per-subroutine check hooks> | |
617 | notes that it can be used to inline a subroutine, by replacing it with a | |
618 | custom op. However there is no further detail of the code needed to do this. | |
619 | It would be useful to add one or more annotated examples of how to create | |
620 | XSUBs that inline. | |
621 | ||
622 | This should provide a measurable speed up to simple XSUBs inside | |
05fb4e20 NC |
623 | tight loops. Initially one would have to write the OP alternative |
624 | implementation by hand, but it's likely that this should be reasonably | |
625 | straightforward for the type of XSUB that would benefit the most. Longer | |
626 | term, once the run-time implementation is proven, it should be possible to | |
627 | progressively update ExtUtils::ParseXS to generate OP implementations for | |
628 | some XSUBs. | |
629 | ||
1ab33739 FC |
630 | =head2 Document how XS modules can install lexical subs |
631 | ||
632 | There is an example in XS::APItest (look for C<lexical_import> in | |
633 | F<ext/XS-APItest/APItest.xs>). The documentation could be based on it. | |
634 | ||
318bf708 NC |
635 | =head2 Remove the use of SVs as temporaries in dump.c |
636 | ||
637 | F<dump.c> contains debugging routines to dump out the contains of perl data | |
638 | structures, such as C<SV>s, C<AV>s and C<HV>s. Currently, the dumping code | |
639 | B<uses> C<SV>s for its temporary buffers, which was a logical initial | |
640 | implementation choice, as they provide ready made memory handling. | |
641 | ||
642 | However, they also lead to a lot of confusion when it happens that what you're | |
643 | trying to debug is seen by the code in F<dump.c>, correctly or incorrectly, as | |
644 | a temporary scalar it can use for a temporary buffer. It's also not possible | |
645 | to dump scalars before the interpreter is properly set up, such as during | |
646 | ithreads cloning. It would be good to progressively replace the use of scalars | |
647 | as string accumulation buffers with something much simpler, directly allocated | |
648 | by C<malloc>. The F<dump.c> code is (or should be) only producing 7 bit | |
649 | US-ASCII, so output character sets are not an issue. | |
650 | ||
651 | Producing and proving an internal simple buffer allocation would make it easier | |
652 | to re-write the internals of the PerlIO subsystem to avoid using C<SV>s for | |
653 | B<its> buffers, use of which can cause problems similar to those of F<dump.c>, | |
654 | at similar times. | |
655 | ||
5d96f598 NC |
656 | =head2 safely supporting POSIX SA_SIGINFO |
657 | ||
658 | Some years ago Jarkko supplied patches to provide support for the POSIX | |
659 | SA_SIGINFO feature in Perl, passing the extra data to the Perl signal handler. | |
660 | ||
661 | Unfortunately, it only works with "unsafe" signals, because under safe | |
662 | signals, by the time Perl gets to run the signal handler, the extra | |
663 | information has been lost. Moreover, it's not easy to store it somewhere, | |
664 | as you can't call mutexs, or do anything else fancy, from inside a signal | |
665 | handler. | |
666 | ||
667 | So it strikes me that we could provide safe SA_SIGINFO support | |
668 | ||
669 | =over 4 | |
670 | ||
671 | =item 1 | |
672 | ||
673 | Provide global variables for two file descriptors | |
674 | ||
675 | =item 2 | |
676 | ||
677 | When the first request is made via C<sigaction> for C<SA_SIGINFO>, create a | |
678 | pipe, store the reader in one, the writer in the other | |
679 | ||
680 | =item 3 | |
681 | ||
682 | In the "safe" signal handler (C<Perl_csighandler()>/C<S_raise_signal()>), if | |
683 | the C<siginfo_t> pointer non-C<NULL>, and the writer file handle is open, | |
684 | ||
685 | =over 8 | |
686 | ||
687 | =item 1 | |
688 | ||
689 | serialise signal number, C<struct siginfo_t> (or at least the parts we care | |
690 | about) into a small auto char buff | |
691 | ||
692 | =item 2 | |
693 | ||
694 | C<write()> that (non-blocking) to the writer fd | |
695 | ||
696 | =over 12 | |
697 | ||
698 | =item 1 | |
699 | ||
700 | if it writes 100%, flag the signal in a counter of "signals on the pipe" akin | |
701 | to the current per-signal-number counts | |
702 | ||
703 | =item 2 | |
704 | ||
705 | if it writes 0%, assume the pipe is full. Flag the data as lost? | |
706 | ||
707 | =item 3 | |
708 | ||
709 | if it writes partially, croak a panic, as your OS is broken. | |
710 | ||
711 | =back | |
712 | ||
713 | =back | |
714 | ||
715 | =item 4 | |
716 | ||
717 | in the regular C<PERL_ASYNC_CHECK()> processing, if there are "signals on | |
718 | the pipe", read the data out, deserialise, build the Perl structures on | |
719 | the stack (code in C<Perl_sighandler()>, the "unsafe" handler), and call as | |
720 | usual. | |
721 | ||
722 | =back | |
723 | ||
724 | I think that this gets us decent C<SA_SIGINFO> support, without the current risk | |
725 | of running Perl code inside the signal handler context. (With all the dangers | |
726 | of things like C<malloc> corruption that that currently offers us) | |
727 | ||
728 | For more information see the thread starting with this message: | |
b4af8972 | 729 | L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-03/msg00305.html> |
5d96f598 | 730 | |
6d71adcd NC |
731 | =head2 autovivification |
732 | ||
733 | Make all autovivification consistent w.r.t LVALUE/RVALUE and strict/no strict; | |
734 | ||
735 | This task is incremental - even a little bit of work on it will help. | |
736 | ||
737 | =head2 Unicode in Filenames | |
738 | ||
739 | chdir, chmod, chown, chroot, exec, glob, link, lstat, mkdir, open, | |
740 | opendir, qx, readdir, readlink, rename, rmdir, stat, symlink, sysopen, | |
741 | system, truncate, unlink, utime, -X. All these could potentially accept | |
742 | Unicode filenames either as input or output (and in the case of system | |
743 | and qx Unicode in general, as input or output to/from the shell). | |
744 | Whether a filesystem - an operating system pair understands Unicode in | |
745 | filenames varies. | |
746 | ||
747 | Known combinations that have some level of understanding include | |
748 | Microsoft NTFS, Apple HFS+ (In Mac OS 9 and X) and Apple UFS (in Mac | |
749 | OS X), NFS v4 is rumored to be Unicode, and of course Plan 9. How to | |
750 | create Unicode filenames, what forms of Unicode are accepted and used | |
751 | (UCS-2, UTF-16, UTF-8), what (if any) is the normalization form used, | |
752 | and so on, varies. Finding the right level of interfacing to Perl | |
753 | requires some thought. Remember that an OS does not implicate a | |
754 | filesystem. | |
755 | ||
756 | (The Windows -C command flag "wide API support" has been at least | |
757 | temporarily retired in 5.8.1, and the -C has been repurposed, see | |
758 | L<perlrun>.) | |
759 | ||
87a942b1 JH |
760 | Most probably the right way to do this would be this: |
761 | L</"Virtualize operating system access">. | |
762 | ||
6d71adcd NC |
763 | =head2 Unicode in %ENV |
764 | ||
765 | Currently the %ENV entries are always byte strings. | |
87a942b1 | 766 | See L</"Virtualize operating system access">. |
6d71adcd | 767 | |
799c141b SH |
768 | (See RT ticket #113536 for information on Win32's handling of %ENV, |
769 | which was fixed to work with native ANSI codepage characters in the | |
770 | environment, but still doesn't work with other characters outside of | |
771 | that codepage present in the environment.) | |
772 | ||
1f2e7916 JD |
773 | =head2 Unicode and glob() |
774 | ||
775 | Currently glob patterns and filenames returned from File::Glob::glob() | |
87a942b1 | 776 | are always byte strings. See L</"Virtualize operating system access">. |
1f2e7916 | 777 | |
6d71adcd NC |
778 | =head2 use less 'memory' |
779 | ||
780 | Investigate trade offs to switch out perl's choices on memory usage. | |
781 | Particularly perl should be able to give memory back. | |
782 | ||
783 | This task is incremental - even a little bit of work on it will help. | |
784 | ||
785 | =head2 Re-implement C<:unique> in a way that is actually thread-safe | |
786 | ||
787 | The old implementation made bad assumptions on several levels. A good 90% | |
788 | solution might be just to make C<:unique> work to share the string buffer | |
789 | of SvPVs. That way large constant strings can be shared between ithreads, | |
790 | such as the configuration information in F<Config>. | |
791 | ||
792 | =head2 Make tainting consistent | |
793 | ||
794 | Tainting would be easier to use if it didn't take documented shortcuts and | |
795 | allow taint to "leak" everywhere within an expression. | |
796 | ||
797 | =head2 readpipe(LIST) | |
798 | ||
799 | system() accepts a LIST syntax (and a PROGRAM LIST syntax) to avoid | |
800 | running a shell. readpipe() (the function behind qx//) could be similarly | |
3b17061e FC |
801 | extended. Note that changing readpipe() itself may not be the solution, as |
802 | it currently has unary precedence, and allowing a list would change the | |
803 | precedence. | |
6d71adcd | 804 | |
6d71adcd NC |
805 | =head2 Audit the code for destruction ordering assumptions |
806 | ||
807 | Change 25773 notes | |
808 | ||
f185f654 KW |
809 | /* Need to check SvMAGICAL, as during global destruction it may be that |
810 | AvARYLEN(av) has been freed before av, and hence the SvANY() pointer | |
811 | is now part of the linked list of SV heads, rather than pointing to | |
812 | the original body. */ | |
813 | /* FIXME - audit the code for other bugs like this one. */ | |
6d71adcd NC |
814 | |
815 | adding the C<SvMAGICAL> check to | |
816 | ||
817 | if (AvARYLEN(av) && SvMAGICAL(AvARYLEN(av))) { | |
818 | MAGIC *mg = mg_find (AvARYLEN(av), PERL_MAGIC_arylen); | |
819 | ||
820 | Go through the core and look for similar assumptions that SVs have particular | |
821 | types, as all bets are off during global destruction. | |
822 | ||
749904bf JH |
823 | =head2 Extend PerlIO and PerlIO::Scalar |
824 | ||
825 | PerlIO::Scalar doesn't know how to truncate(). Implementing this | |
826 | would require extending the PerlIO vtable. | |
827 | ||
828 | Similarly the PerlIO vtable doesn't know about formats (write()), or | |
829 | about stat(), or chmod()/chown(), utime(), or flock(). | |
830 | ||
831 | (For PerlIO::Scalar it's hard to see what e.g. mode bits or ownership | |
832 | would mean.) | |
833 | ||
834 | PerlIO doesn't do directories or symlinks, either: mkdir(), rmdir(), | |
835 | opendir(), closedir(), seekdir(), rewinddir(), glob(); symlink(), | |
836 | readlink(). | |
837 | ||
94da6c29 JH |
838 | See also L</"Virtualize operating system access">. |
839 | ||
d6c1e11f JH |
840 | =head2 Organize error messages |
841 | ||
842 | Perl's diagnostics (error messages, see L<perldiag>) could use | |
a8d0aeb9 | 843 | reorganizing and formalizing so that each error message has its |
d6c1e11f JH |
844 | stable-for-all-eternity unique id, categorized by severity, type, and |
845 | subsystem. (The error messages would be listed in a datafile outside | |
c4bd451b CB |
846 | of the Perl source code, and the source code would only refer to the |
847 | messages by the id.) This clean-up and regularizing should apply | |
d6c1e11f JH |
848 | for all croak() messages. |
849 | ||
850 | This would enable all sorts of things: easier translation/localization | |
851 | of the messages (though please do keep in mind the caveats of | |
852 | L<Locale::Maketext> about too straightforward approaches to | |
853 | translation), filtering by severity, and instead of grepping for a | |
854 | particular error message one could look for a stable error id. (Of | |
855 | course, changing the error messages by default would break all the | |
856 | existing software depending on some particular error message...) | |
857 | ||
858 | This kind of functionality is known as I<message catalogs>. Look for | |
859 | inspiration for example in the catgets() system, possibly even use it | |
860 | if available-- but B<only> if available, all platforms will B<not> | |
de96509d | 861 | have catgets(). |
d6c1e11f JH |
862 | |
863 | For the really pure at heart, consider extending this item to cover | |
47a9c258 | 864 | also the warning messages (see L<warnings>, F<regen/warnings.pl>). |
3236f110 | 865 | |
0bdfc961 | 866 | =head1 Tasks that need a knowledge of the interpreter |
3298bd4d | 867 | |
0bdfc961 NC |
868 | These tasks would need C knowledge, and knowledge of how the interpreter works, |
869 | or a willingness to learn. | |
3298bd4d | 870 | |
514e62e3 FC |
871 | =head2 fix refaliasing with nested and recursive subroutines |
872 | ||
873 | Currently aliasing lexical variables via reference only applies to the | |
874 | current subroutine, and does not propagate to inner closures, nor does | |
875 | aliasing of outer variables within closures propagate to the outer | |
876 | subroutine. This is because each subroutine has its own lexical pad and the | |
877 | aliasing works by changing which SV the pad points to. | |
878 | ||
879 | One possible way to fix this would be to create new ops for accessing | |
880 | variables that are closed over. So C<my $x; sub {$x}> would use a new op | |
881 | type, say C<padoutsv>, instead of the C<padsv> currently used in the | |
882 | sub. That new op would possibly check a flag or some such and see if it | |
883 | needs to fetch the variable from an outer pad. If we follow this approach, | |
884 | it should be possible at compile time to detect cases where the more | |
885 | complex C<padoutsv> op is unnecessary and revert back to the simpler, | |
886 | faster C<padsv>. There would need to be corresponding ops for arrays, | |
887 | hashes, and subs, too. | |
888 | ||
889 | There is also a related issue with recursion and C<state> variables. A | |
890 | subroutine actually has a list of lexical pads, each one used at a | |
891 | different recursion level. If a C<state> variable is aliased to another | |
892 | variable after a recursive call to the same subroutine, that higher call | |
893 | depth will not see the effect of aliasing, because the second pad will have | |
894 | been created already. Similarly, aliasing a state variable within a | |
895 | recursive call will not affect outer calls, even though all call depths are | |
896 | supposed to share the same C<state> variables. | |
897 | ||
898 | Both of these bugs affect C<foreach> aliasing, too. | |
899 | ||
10517af5 JD |
900 | =head2 forbid labels with keyword names |
901 | ||
902 | Currently C<goto keyword> "computes" the label value: | |
903 | ||
904 | $ perl -e 'goto print' | |
905 | Can't find label 1 at -e line 1. | |
906 | ||
343c8006 JD |
907 | It is controversial if the right way to avoid the confusion is to forbid |
908 | labels with keyword names, or if it would be better to always treat | |
909 | bareword expressions after a "goto" as a label and never as a keyword. | |
10517af5 | 910 | |
de6375e3 RGS |
911 | =head2 truncate() prototype |
912 | ||
913 | The prototype of truncate() is currently C<$$>. It should probably | |
f3fccad6 | 914 | be C<*$> instead. (This is changed in F<regen/opcodes>.) |
de6375e3 | 915 | |
565590b5 NC |
916 | =head2 error reporting of [$a ; $b] |
917 | ||
918 | Using C<;> inside brackets is a syntax error, and we don't propose to change | |
919 | that by giving it any meaning. However, it's not reported very helpfully: | |
920 | ||
921 | $ perl -e '$a = [$b; $c];' | |
922 | syntax error at -e line 1, near "$b;" | |
923 | syntax error at -e line 1, near "$c]" | |
924 | Execution of -e aborted due to compilation errors. | |
925 | ||
926 | It should be possible to hook into the tokeniser or the lexer, so that when a | |
927 | C<;> is parsed where it is not legal as a statement terminator (ie inside | |
928 | C<{}> used as a hashref, C<[]> or C<()>) it issues an error something like | |
929 | I<';' isn't legal inside an expression - if you need multiple statements use a | |
930 | do {...} block>. See the thread starting at | |
b4af8972 | 931 | L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-09/msg00573.html> |
565590b5 | 932 | |
e053a921 RS |
933 | =head2 strict as warnings |
934 | ||
b8b41556 | 935 | See L<http://markmail.org/message/vbrupaslr3bybmvk>, where Joshua ben Jore |
e053a921 RS |
936 | writes: I've been of the opinion that everything strict.pm does ought to be |
937 | able to considered just warnings that have been promoted to 'FATAL'. | |
938 | ||
718140ec NC |
939 | =head2 lexicals used only once |
940 | ||
941 | This warns: | |
942 | ||
943 | $ perl -we '$pie = 42' | |
944 | Name "main::pie" used only once: possible typo at -e line 1. | |
945 | ||
946 | This does not: | |
947 | ||
948 | $ perl -we 'my $pie = 42' | |
949 | ||
950 | Logically all lexicals used only once should warn, if the user asks for | |
d6f4ea2e SP |
951 | warnings. An unworked RT ticket (#5087) has been open for almost seven |
952 | years for this discrepancy. | |
718140ec | 953 | |
a3d15f9a RGS |
954 | =head2 UTF-8 revamp |
955 | ||
85c006b6 KW |
956 | The handling of Unicode is unclean in many places. In the regex engine |
957 | there are especially many problems. The swash data structure could be | |
958 | replaced my something better. Inversion lists and maps are likely | |
959 | candidates. The whole Unicode database could be placed in-core for a | |
960 | huge speed-up. Only minimal work was done on the optimizer when utf8 | |
961 | was added, with the result that the synthetic start class often will | |
962 | fail to narrow down the possible choices when given non-Latin1 input. | |
4e1c9055 | 963 | Karl Williamson has been working on this - talk to him. |
a3d15f9a | 964 | |
636e63cb NC |
965 | =head2 state variable initialization in list context |
966 | ||
967 | Currently this is illegal: | |
968 | ||
969 | state ($a, $b) = foo(); | |
970 | ||
a2874905 | 971 | In Perl 6, C<state ($a) = foo();> and C<(state $a) = foo();> have different |
a8d0aeb9 | 972 | semantics, which is tricky to implement in Perl 5 as currently they produce |
a2874905 | 973 | the same opcode trees. The Perl 6 design is firm, so it would be good to |
a8d0aeb9 | 974 | implement the necessary code in Perl 5. There are comments in |
a2874905 NC |
975 | C<Perl_newASSIGNOP()> that show the code paths taken by various assignment |
976 | constructions involving state variables. | |
636e63cb | 977 | |
a393eb28 RGS |
978 | =head2 A does() built-in |
979 | ||
980 | Like ref(), only useful. It would call the C<DOES> method on objects; it | |
981 | would also tell whether something can be dereferenced as an | |
982 | array/hash/etc., or used as a regexp, etc. | |
983 | L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-03/msg00481.html> | |
984 | ||
985 | =head2 Tied filehandles and write() don't mix | |
986 | ||
987 | There is no method on tied filehandles to allow them to be called back by | |
988 | formats. | |
4fedb12c | 989 | |
53967bb9 RGS |
990 | =head2 Propagate compilation hints to the debugger |
991 | ||
992 | Currently a debugger started with -dE on the command-line doesn't see the | |
993 | features enabled by -E. More generally hints (C<$^H> and C<%^H>) aren't | |
994 | propagated to the debugger. Probably it would be a good thing to propagate | |
995 | hints from the innermost non-C<DB::> scope: this would make code eval'ed | |
996 | in the debugger see the features (and strictures, etc.) currently in | |
997 | scope. | |
998 | ||
d10fc472 | 999 | =head2 Attach/detach debugger from running program |
1626a787 | 1000 | |
cd793d32 NC |
1001 | The old perltodo notes "With C<gdb>, you can attach the debugger to a running |
1002 | program if you pass the process ID. It would be good to do this with the Perl | |
0bdfc961 NC |
1003 | debugger on a running Perl program, although I'm not sure how it would be |
1004 | done." ssh and screen do this with named pipes in /tmp. Maybe we can too. | |
1626a787 | 1005 | |
c5fb089a | 1006 | =head2 regexp optimizer optional |
0bdfc961 | 1007 | |
c5fb089a DS |
1008 | The regexp optimizer is not optional. It should be configurable to be optional |
1009 | and to allow its performance to be measured and its bugs to be easily | |
1010 | demonstrated. | |
0bdfc961 | 1011 | |
ef36c6a7 RGS |
1012 | =head2 C</w> regex modifier |
1013 | ||
1014 | That flag would enable to match whole words, and also to interpolate | |
1015 | arrays as alternations. With it, C</P/w> would be roughly equivalent to: | |
1016 | ||
1017 | do { local $"='|'; /\b(?:P)\b/ } | |
1018 | ||
b4af8972 RB |
1019 | See |
1020 | L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-01/msg00400.html> | |
ef36c6a7 RGS |
1021 | for the discussion. |
1022 | ||
0bdfc961 NC |
1023 | =head2 optional optimizer |
1024 | ||
1025 | Make the peephole optimizer optional. Currently it performs two tasks as | |
1026 | it walks the optree - genuine peephole optimisations, and necessary fixups of | |
1027 | ops. It would be good to find an efficient way to switch out the | |
1028 | optimisations whilst keeping the fixups. | |
1029 | ||
1030 | =head2 You WANT *how* many | |
1031 | ||
1032 | Currently contexts are void, scalar and list. split has a special mechanism in | |
1033 | place to pass in the number of return values wanted. It would be useful to | |
1034 | have a general mechanism for this, backwards compatible and little speed hit. | |
1035 | This would allow proposals such as short circuiting sort to be implemented | |
1036 | as a module on CPAN. | |
1037 | ||
de535794 | 1038 | =head2 Self-ties |
2810d901 | 1039 | |
de535794 | 1040 | Self-ties are currently illegal because they caused too many segfaults. Maybe |
a8d0aeb9 | 1041 | the causes of these could be tracked down and self-ties on all types |
de535794 | 1042 | reinstated. |
0bdfc961 NC |
1043 | |
1044 | =head2 Optimize away @_ | |
1045 | ||
1046 | The old perltodo notes "Look at the "reification" code in C<av.c>". | |
1047 | ||
87a942b1 JH |
1048 | =head2 Virtualize operating system access |
1049 | ||
1050 | Implement a set of "vtables" that virtualizes operating system access | |
9fe0b8be RS |
1051 | (chdir(), chmod(), dbmopen(), getenv(), glob(), link(), mkdir(), open(), |
1052 | opendir(), readdir(), rename(), rmdir(), stat(), sysopen(), uname(), | |
1053 | unlink(), etc.) At the very least these interfaces should take SVs as | |
1054 | "name" arguments instead of bare char pointers; probably the most | |
1055 | flexible and extensible way would be for the Perl-facing interfaces to | |
1056 | accept HVs. The system needs to be per-operating-system and | |
1057 | per-file-system hookable/filterable, preferably both from XS and Perl | |
1058 | level (L<perlport/"Files and Filesystems"> is good reading at this | |
1059 | point, in fact, all of L<perlport> is.) | |
87a942b1 | 1060 | |
e1a3d5d1 JH |
1061 | This has actually already been implemented (but only for Win32), |
1062 | take a look at F<iperlsys.h> and F<win32/perlhost.h>. While all Win32 | |
1063 | variants go through a set of "vtables" for operating system access, | |
e1020413 | 1064 | non-Win32 systems currently go straight for the POSIX/Unix-style |
e1a3d5d1 JH |
1065 | system/library call. Similar system as for Win32 should be |
1066 | implemented for all platforms. The existing Win32 implementation | |
1067 | probably does not need to survive alongside this proposed new | |
1068 | implementation, the approaches could be merged. | |
87a942b1 JH |
1069 | |
1070 | What would this give us? One often-asked-for feature this would | |
94da6c29 JH |
1071 | enable is using Unicode for filenames, and other "names" like %ENV, |
1072 | usernames, hostnames, and so forth. | |
1073 | (See L<perlunicode/"When Unicode Does Not Happen">.) | |
1074 | ||
1075 | But this kind of virtualization would also allow for things like | |
1076 | virtual filesystems, virtual networks, and "sandboxes" (though as long | |
1077 | as dynamic loading of random object code is allowed, not very safe | |
1078 | sandboxes since external code of course know not of Perl's vtables). | |
1079 | An example of a smaller "sandbox" is that this feature can be used to | |
1080 | implement per-thread working directories: Win32 already does this. | |
1081 | ||
1082 | See also L</"Extend PerlIO and PerlIO::Scalar">. | |
87a942b1 | 1083 | |
52960e22 JC |
1084 | =head2 repack the optree |
1085 | ||
af4a745c FC |
1086 | B<Note:> This entry was written in reference to the I<old> slab allocator, |
1087 | removed in commit 7aef8e5bd14. | |
1088 | ||
52960e22 | 1089 | Repacking the optree after execution order is determined could allow |
057163d7 | 1090 | removal of NULL ops, and optimal ordering of OPs with respect to cache-line |
2723c0fb | 1091 | filling. I think that |
057163d7 NC |
1092 | the best way to do this is to make it an optional step just before the |
1093 | completed optree is attached to anything else, and to use the slab allocator | |
2723c0fb FC |
1094 | unchanged--but allocate a single slab the right size, avoiding partial |
1095 | slabs--, so that freeing ops is identical whether or not this step runs. | |
057163d7 NC |
1096 | Note that the slab allocator allocates ops downwards in memory, so one would |
1097 | have to actually "allocate" the ops in reverse-execution order to get them | |
1098 | contiguous in memory in execution order. | |
1099 | ||
b4af8972 RB |
1100 | See |
1101 | L<http://www.nntp.perl.org/group/perl.perl5.porters/2007/12/msg131975.html> | |
057163d7 NC |
1102 | |
1103 | Note that running this copy, and then freeing all the old location ops would | |
1104 | cause their slabs to be freed, which would eliminate possible memory wastage if | |
1105 | the previous suggestion is implemented, and we swap slabs more frequently. | |
52960e22 | 1106 | |
12e06b6f NC |
1107 | =head2 eliminate incorrect line numbers in warnings |
1108 | ||
1109 | This code | |
1110 | ||
1111 | use warnings; | |
1112 | my $undef; | |
f703fc96 | 1113 | |
12e06b6f NC |
1114 | if ($undef == 3) { |
1115 | } elsif ($undef == 0) { | |
1116 | } | |
1117 | ||
18a16cc5 | 1118 | used to produce this output: |
12e06b6f NC |
1119 | |
1120 | Use of uninitialized value in numeric eq (==) at wrong.pl line 4. | |
1121 | Use of uninitialized value in numeric eq (==) at wrong.pl line 4. | |
1122 | ||
18a16cc5 NC |
1123 | where the line of the second warning was misreported - it should be line 5. |
1124 | Rafael fixed this - the problem arose because there was no nextstate OP | |
1125 | between the execution of the C<if> and the C<elsif>, hence C<PL_curcop> still | |
1126 | reports that the currently executing line is line 4. The solution was to inject | |
1127 | a nextstate OPs for each C<elsif>, although it turned out that the nextstate | |
1128 | OP needed to be a nulled OP, rather than a live nextstate OP, else other line | |
1129 | numbers became misreported. (Jenga!) | |
12e06b6f NC |
1130 | |
1131 | The problem is more general than C<elsif> (although the C<elsif> case is the | |
1132 | most common and the most confusing). Ideally this code | |
1133 | ||
1134 | use warnings; | |
1135 | my $undef; | |
f703fc96 | 1136 | |
12e06b6f NC |
1137 | my $a = $undef + 1; |
1138 | my $b | |
1139 | = $undef | |
1140 | + 1; | |
1141 | ||
1142 | would produce this output | |
1143 | ||
f185f654 KW |
1144 | Use of uninitialized value $undef in addition (+) at wrong.pl line 4. |
1145 | Use of uninitialized value $undef in addition (+) at wrong.pl line 7. | |
12e06b6f NC |
1146 | |
1147 | (rather than lines 4 and 5), but this would seem to require every OP to carry | |
1148 | (at least) line number information. | |
1149 | ||
1150 | What might work is to have an optional line number in memory just before the | |
1151 | BASEOP structure, with a flag bit in the op to say whether it's present. | |
1152 | Initially during compile every OP would carry its line number. Then add a late | |
c5fb089a | 1153 | pass to the optimizer (potentially combined with L</repack the optree>) which |
12e06b6f NC |
1154 | looks at the two ops on every edge of the graph of the execution path. If |
1155 | the line number changes, flags the destination OP with this information. | |
1156 | Once all paths are traced, replace every op with the flag with a | |
1157 | nextstate-light op (that just updates C<PL_curcop>), which in turn then passes | |
1158 | control on to the true op. All ops would then be replaced by variants that | |
1159 | do not store the line number. (Which, logically, why it would work best in | |
1160 | conjunction with L</repack the optree>, as that is already copying/reallocating | |
1161 | all the OPs) | |
1162 | ||
18a16cc5 NC |
1163 | (Although I should note that we're not certain that doing this for the general |
1164 | case is worth it) | |
1165 | ||
52960e22 JC |
1166 | =head2 optimize tail-calls |
1167 | ||
1168 | Tail-calls present an opportunity for broadly applicable optimization; | |
1169 | anywhere that C<< return foo(...) >> is called, the outer return can | |
1170 | be replaced by a goto, and foo will return directly to the outer | |
1171 | caller, saving (conservatively) 25% of perl's call&return cost, which | |
1172 | is relatively higher than in C. The scheme language is known to do | |
1173 | this heavily. B::Concise provides good insight into where this | |
1174 | optimization is possible, ie anywhere entersub,leavesub op-sequence | |
1175 | occurs. | |
1176 | ||
1177 | perl -MO=Concise,-exec,a,b,-main -e 'sub a{ 1 }; sub b {a()}; b(2)' | |
1178 | ||
1179 | Bottom line on this is probably a new pp_tailcall function which | |
1180 | combines the code in pp_entersub, pp_leavesub. This should probably | |
1181 | be done 1st in XS, and using B::Generate to patch the new OP into the | |
1182 | optrees. | |
1183 | ||
e12cb30b | 1184 | =head2 Add C<0odddd> |
0c397127 KW |
1185 | |
1186 | It has been proposed that octal constants be specifiable through the syntax | |
1187 | C<0oddddd>, parallel to the existing construct to specify hex constants | |
1188 | C<0xddddd> | |
1189 | ||
bf7d9bd8 AC |
1190 | =head2 Revisit the regex super-linear cache code |
1191 | ||
1192 | Perl executes regexes using the traditional backtracking algorithm, which | |
1193 | makes it possible to implement a variety of powerful pattern-matching | |
1194 | features (like embedded code blocks), at the cost of taking exponential time | |
1195 | to run on some pathological patterns. The exponential-time problem is | |
1196 | mitigated by the I<super-linear cache>, which detects when we're processing | |
1197 | such a pathological pattern, and does some additional bookkeeping to avoid | |
1198 | much of the work. However, that code has bit-rotted a little; some patterns | |
1199 | don't make as much use of it as they should. The proposal is to analyse | |
1200 | where the current cache code has problems, and extend it to cover those cases. | |
1201 | ||
1202 | See also | |
1203 | L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2013-01/msg00339.html> | |
1204 | ||
0bdfc961 NC |
1205 | =head1 Big projects |
1206 | ||
1207 | Tasks that will get your name mentioned in the description of the "Highlights | |
54a814f4 | 1208 | of 5.26.0" |
0bdfc961 NC |
1209 | |
1210 | =head2 make ithreads more robust | |
1211 | ||
45a81a90 | 1212 | Generally make ithreads more robust. |
0bdfc961 NC |
1213 | |
1214 | This task is incremental - even a little bit of work on it will help, and | |
1215 | will be greatly appreciated. | |
1216 | ||
07577ec1 FC |
1217 | One bit would be to determine how to clone directory handles on systems |
1218 | without a C<fchdir> function (in sv.c:Perl_dirp_dup). | |
6c047da7 | 1219 | |
59c7f7d5 RGS |
1220 | Fix Perl_sv_dup, et al so that threads can return objects. |
1221 | ||
44a7a252 JV |
1222 | =head1 Tasks for microperl |
1223 | ||
1224 | ||
1225 | [ Each and every one of these may be obsolete, but they were listed | |
1226 | in the old Todo.micro file] | |
1227 | ||
44a7a252 JV |
1228 | =head2 do away with fork/exec/wait? |
1229 | ||
1230 | (system, popen should be enough?) | |
1231 | ||
1232 | =head2 some of the uconfig.sh really needs to be probed (using cc) in buildtime: | |
1233 | ||
1234 | (uConfigure? :-) native datatype widths and endianness come to mind | |
1235 |