This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
perltodo: Revise utf8 todo
[perl5.git] / pod / perltodo.pod
CommitLineData
7711098a
GS
1=head1 NAME
2
3perltodo - Perl TO-DO List
4
5=head1 DESCRIPTION
e50bb9a1 6
049aabcb
NC
7This is a list of wishes for Perl. The most up to date version of this file
8is at http://perl5.git.perl.org/perl.git/blob_plain/HEAD:/pod/perltodo.pod
9
10The tasks we think are smaller or easier are listed first. Anyone is welcome
11to work on any of these, but it's a good idea to first contact
12I<perl5-porters@perl.org> to avoid duplication of effort, and to learn from
13any previous attempts. By all means contact a pumpking privately first if you
14prefer.
e50bb9a1 15
0bdfc961
NC
16Whilst patches to make the list shorter are most welcome, ideas to add to
17the list are also encouraged. Check the perl5-porters archives for past
18ideas, and any discussion about them. One set of archives may be found at:
e50bb9a1 19
0bdfc961 20 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/
938c8732 21
617eabfa
NC
22What can we offer you in return? Fame, fortune, and everlasting glory? Maybe
23not, but if your patch is incorporated, then we'll add your name to the
24F<AUTHORS> file, which ships in the official distribution. How many other
25programming languages offer you 1 line of immortality?
938c8732 26
0bdfc961 27=head1 Tasks that only need Perl knowledge
e50bb9a1 28
de2b17d8
NC
29=head2 Migrate t/ from custom TAP generation
30
31Many tests below F<t/> still generate TAP by "hand", rather than using library
32functions. As explained in L<perlhack/Writing a test>, tests in F<t/> are
33written in a particular way to test that more complex constructions actually
34work before using them routinely. Hence they don't use C<Test::More>, but
35instead there is an intentionally simpler library, F<t/test.pl>. However,
36quite a few tests in F<t/> have not been refactored to use it. Refactoring
37any of these tests, one at a time, is a useful thing TODO.
38
0d8e5a42
RGS
39The subdirectories F<base>, F<cmd> and F<comp>, that contain the most
40basic tests, should be excluded from this task.
41
08063926
NC
42=head2 Test that regen.pl was run
43
44There are various generated files shipped with the perl distribution, for
45things like header files generate from data. The generation scripts are
46written in perl, and all can be run by F<regen.pl>. However, because they're
47written in perl, we can't run them before we've built perl. We can't run them
48as part of the F<Makefile>, because changing files underneath F<make> confuses
49it completely, and we don't want to run them automatically anyway, as they
50change files shipped by the distribution, something we seek not do to.
51
52If someone changes the data, but forgets to re-run F<regen.pl> then the
53generated files are out of sync. It would be good to have a test in
54F<t/porting> that checks that the generated files are in sync, and fails
55otherwise, to alert someone before they make a poor commit. I suspect that this
56would require adapting the scripts run from F<regen.pl> to have dry-run
57options, and invoking them with these, or by refactoring them into a library
58that does the generation, which can be called by the scripts, and by the test.
59
0be987a2
NC
60=head2 Automate perldelta generation
61
62The perldelta file accompanying each release summaries the major changes.
63It's mostly manually generated currently, but some of that could be
64automated with a bit of perl, specifically the generation of
65
66=over
67
68=item Modules and Pragmata
69
70=item New Documentation
71
72=item New Tests
73
74=back
75
76See F<Porting/how_to_write_a_perldelta.pod> for details.
77
5a176cbc
NC
78=head2 Remove duplication of test setup.
79
80Schwern notes, that there's duplication of code - lots and lots of tests have
81some variation on the big block of C<$Is_Foo> checks. We can safely put this
82into a file, change it to build an C<%Is> hash and require it. Maybe just put
83it into F<test.pl>. Throw in the handy tainting subroutines.
84
87a942b1 85=head2 POD -E<gt> HTML conversion in the core still sucks
e50bb9a1 86
938c8732 87Which is crazy given just how simple POD purports to be, and how simple HTML
adebf063
NC
88can be. It's not actually I<as> simple as it sounds, particularly with the
89flexibility POD allows for C<=item>, but it would be good to improve the
90visual appeal of the HTML generated, and to avoid it having any validation
91errors. See also L</make HTML install work>, as the layout of installation tree
92is needed to improve the cross-linking.
938c8732 93
dc0fb092
SP
94The addition of C<Pod::Simple> and its related modules may make this task
95easier to complete.
96
0befdfba
NC
97=head2 Make ExtUtils::ParseXS use strict;
98
99F<lib/ExtUtils/ParseXS.pm> contains this line
100
101 # use strict; # One of these days...
102
103Simply uncomment it, and fix all the resulting issues :-)
104
105The more practical approach, to break the task down into manageable chunks, is
106to work your way though the code from bottom to top, or if necessary adding
107extra C<{ ... }> blocks, and turning on strict within them.
108
0bdfc961 109=head2 Make Schwern poorer
e50bb9a1 110
613bd4f7 111We should have tests for everything. When all the core's modules are tested,
0bdfc961
NC
112Schwern has promised to donate to $500 to TPF. We may need volunteers to
113hold him upside down and shake vigorously in order to actually extract the
114cash.
3958b146 115
0bdfc961 116=head2 Improve the coverage of the core tests
e50bb9a1 117
e1020413 118Use Devel::Cover to ascertain the core modules' test coverage, then add
02f21748 119tests that are currently missing.
30222c0f 120
0bdfc961 121=head2 test B
e50bb9a1 122
0bdfc961 123A full test suite for the B module would be nice.
e50bb9a1 124
0bdfc961 125=head2 A decent benchmark
e50bb9a1 126
617eabfa 127C<perlbench> seems impervious to any recent changes made to the perl core. It
0bdfc961
NC
128would be useful to have a reasonable general benchmarking suite that roughly
129represented what current perl programs do, and measurably reported whether
130tweaks to the core improve, degrade or don't really affect performance, to
131guide people attempting to optimise the guts of perl. Gisle would welcome
132new tests for perlbench.
6168cf99 133
0bdfc961 134=head2 fix tainting bugs
6168cf99 135
0bdfc961
NC
136Fix the bugs revealed by running the test suite with the C<-t> switch (via
137C<make test.taintwarn>).
e50bb9a1 138
0bdfc961 139=head2 Dual life everything
e50bb9a1 140
0bdfc961
NC
141As part of the "dists" plan, anything that doesn't belong in the smallest perl
142distribution needs to be dual lifed. Anything else can be too. Figure out what
143changes would be needed to package that module and its tests up for CPAN, and
144do so. Test it with older perl releases, and fix the problems you find.
e50bb9a1 145
a393eb28
RGS
146To make a minimal perl distribution, it's useful to look at
147F<t/lib/commonsense.t>.
148
0bdfc961 149=head2 POSIX memory footprint
e50bb9a1 150
0bdfc961
NC
151Ilya observed that use POSIX; eats memory like there's no tomorrow, and at
152various times worked to cut it down. There is probably still fat to cut out -
153for example POSIX passes Exporter some very memory hungry data structures.
e50bb9a1 154
eed36644
NC
155=head2 embed.pl/makedef.pl
156
157There is a script F<embed.pl> that generates several header files to prefix
158all of Perl's symbols in a consistent way, to provide some semblance of
159namespace support in C<C>. Functions are declared in F<embed.fnc>, variables
907b3e23 160in F<interpvar.h>. Quite a few of the functions and variables
eed36644
NC
161are conditionally declared there, using C<#ifdef>. However, F<embed.pl>
162doesn't understand the C macros, so the rules about which symbols are present
163when is duplicated in F<makedef.pl>. Writing things twice is bad, m'kay.
164It would be good to teach C<embed.pl> to understand the conditional
165compilation, and hence remove the duplication, and the mistakes it has caused.
e50bb9a1 166
801de10e
NC
167=head2 use strict; and AutoLoad
168
169Currently if you write
170
171 package Whack;
172 use AutoLoader 'AUTOLOAD';
173 use strict;
174 1;
175 __END__
176 sub bloop {
177 print join (' ', No, strict, here), "!\n";
178 }
179
180then C<use strict;> isn't in force within the autoloaded subroutines. It would
181be more consistent (and less surprising) to arrange for all lexical pragmas
182in force at the __END__ block to be in force within each autoloaded subroutine.
183
773b3597
RGS
184There's a similar problem with SelfLoader.
185
91d0cbf6
NC
186=head2 profile installman
187
188The F<installman> script is slow. All it is doing text processing, which we're
189told is something Perl is good at. So it would be nice to know what it is doing
190that is taking so much CPU, and where possible address it.
191
a9ed9b74
JV
192=head2 enable lexical enabling/disabling of inidvidual warnings
193
194Currently, warnings can only be enabled or disabled by category. There
195are times when it would be useful to quash a single warning, not a
196whole category.
91d0cbf6 197
0bdfc961 198=head1 Tasks that need a little sysadmin-type knowledge
e50bb9a1 199
0bdfc961
NC
200Or if you prefer, tasks that you would learn from, and broaden your skills
201base...
e50bb9a1 202
cd793d32 203=head2 make HTML install work
e50bb9a1 204
adebf063
NC
205There is an C<installhtml> target in the Makefile. It's marked as
206"experimental". It would be good to get this tested, make it work reliably, and
207remove the "experimental" tag. This would include
208
209=over 4
210
211=item 1
212
213Checking that cross linking between various parts of the documentation works.
214In particular that links work between the modules (files with POD in F<lib/>)
215and the core documentation (files in F<pod/>)
216
217=item 2
218
617eabfa
NC
219Work out how to split C<perlfunc> into chunks, preferably one per function
220group, preferably with general case code that could be used elsewhere.
221Challenges here are correctly identifying the groups of functions that go
222together, and making the right named external cross-links point to the right
223page. Things to be aware of are C<-X>, groups such as C<getpwnam> to
224C<endservent>, two or more C<=items> giving the different parameter lists, such
225as
adebf063
NC
226
227 =item substr EXPR,OFFSET,LENGTH,REPLACEMENT
adebf063 228 =item substr EXPR,OFFSET,LENGTH
adebf063
NC
229 =item substr EXPR,OFFSET
230
231and different parameter lists having different meanings. (eg C<select>)
232
233=back
3a89a73c 234
0bdfc961
NC
235=head2 compressed man pages
236
237Be able to install them. This would probably need a configure test to see how
238the system does compressed man pages (same directory/different directory?
239same filename/different filename), as well as tweaking the F<installman> script
240to compress as necessary.
241
30222c0f
NC
242=head2 Add a code coverage target to the Makefile
243
244Make it easy for anyone to run Devel::Cover on the core's tests. The steps
245to do this manually are roughly
246
247=over 4
248
249=item *
250
251do a normal C<Configure>, but include Devel::Cover as a module to install
f11a3063 252(see L<INSTALL> for how to do this)
30222c0f
NC
253
254=item *
255
256 make perl
257
258=item *
259
260 cd t; HARNESS_PERL_SWITCHES=-MDevel::Cover ./perl -I../lib harness
261
262=item *
263
264Process the resulting Devel::Cover database
265
266=back
267
268This just give you the coverage of the F<.pm>s. To also get the C level
269coverage you need to
270
271=over 4
272
273=item *
274
275Additionally tell C<Configure> to use the appropriate C compiler flags for
276C<gcov>
277
278=item *
279
280 make perl.gcov
281
282(instead of C<make perl>)
283
284=item *
285
286After running the tests run C<gcov> to generate all the F<.gcov> files.
287(Including down in the subdirectories of F<ext/>
288
289=item *
290
291(From the top level perl directory) run C<gcov2perl> on all the C<.gcov> files
292to get their stats into the cover_db directory.
293
294=item *
295
296Then process the Devel::Cover database
297
298=back
299
300It would be good to add a single switch to C<Configure> to specify that you
301wanted to perform perl level coverage, and another to specify C level
302coverage, and have C<Configure> and the F<Makefile> do all the right things
303automatically.
304
02f21748 305=head2 Make Config.pm cope with differences between built and installed perl
0bdfc961
NC
306
307Quite often vendors ship a perl binary compiled with their (pay-for)
308compilers. People install a free compiler, such as gcc. To work out how to
309build extensions, Perl interrogates C<%Config>, so in this situation
310C<%Config> describes compilers that aren't there, and extension building
311fails. This forces people into choosing between re-compiling perl themselves
312using the compiler they have, or only using modules that the vendor ships.
313
314It would be good to find a way teach C<Config.pm> about the installation setup,
315possibly involving probing at install time or later, so that the C<%Config> in
316a binary distribution better describes the installed machine, when the
317installed machine differs from the build machine in some significant way.
318
728f4ecd
NC
319=head2 linker specification files
320
321Some platforms mandate that you provide a list of a shared library's external
322symbols to the linker, so the core already has the infrastructure in place to
323do this for generating shared perl libraries. My understanding is that the
324GNU toolchain can accept an optional linker specification file, and restrict
325visibility just to symbols declared in that file. It would be good to extend
326F<makedef.pl> to support this format, and to provide a means within
327C<Configure> to enable it. This would allow Unix users to test that the
328export list is correct, and to build a perl that does not pollute the global
32d539f5
RU
329namespace with private symbols, and will fail in the same way as msvc or mingw
330builds or when using PERL_DL_NONLAZY=1.
728f4ecd 331
a229ae3b
RGS
332=head2 Cross-compile support
333
334Currently C<Configure> understands C<-Dusecrosscompile> option. This option
335arranges for building C<miniperl> for TARGET machine, so this C<miniperl> is
336assumed then to be copied to TARGET machine and used as a replacement of full
337C<perl> executable.
338
d1307786 339This could be done little differently. Namely C<miniperl> should be built for
a229ae3b 340HOST and then full C<perl> with extensions should be compiled for TARGET.
d1307786 341This, however, might require extra trickery for %Config: we have one config
87a942b1
JH
342first for HOST and then another for TARGET. Tools like MakeMaker will be
343mightily confused. Having around two different types of executables and
344libraries (HOST and TARGET) makes life interesting for Makefiles and
345shell (and Perl) scripts. There is $Config{run}, normally empty, which
346can be used as an execution wrapper. Also note that in some
347cross-compilation/execution environments the HOST and the TARGET do
348not see the same filesystem(s), the $Config{run} may need to do some
349file/directory copying back and forth.
0bdfc961 350
8537f021
RGS
351=head2 roffitall
352
353Make F<pod/roffitall> be updated by F<pod/buildtoc>.
354
98fca0e8
NC
355=head2 Split "linker" from "compiler"
356
357Right now, Configure probes for two commands, and sets two variables:
358
359=over 4
360
b91dd380 361=item * C<cc> (in F<cc.U>)
98fca0e8
NC
362
363This variable holds the name of a command to execute a C compiler which
364can resolve multiple global references that happen to have the same
365name. Usual values are F<cc> and F<gcc>.
366Fervent ANSI compilers may be called F<c89>. AIX has F<xlc>.
367
b91dd380 368=item * C<ld> (in F<dlsrc.U>)
98fca0e8
NC
369
370This variable indicates the program to be used to link
371libraries for dynamic loading. On some systems, it is F<ld>.
372On ELF systems, it should be C<$cc>. Mostly, we'll try to respect
373the hint file setting.
374
375=back
376
8d159ec1
NC
377There is an implicit historical assumption from around Perl5.000alpha
378something, that C<$cc> is also the correct command for linking object files
379together to make an executable. This may be true on Unix, but it's not true
380on other platforms, and there are a maze of work arounds in other places (such
381as F<Makefile.SH>) to cope with this.
98fca0e8
NC
382
383Ideally, we should create a new variable to hold the name of the executable
384linker program, probe for it in F<Configure>, and centralise all the special
385case logic there or in hints files.
386
387A small bikeshed issue remains - what to call it, given that C<$ld> is already
8d159ec1
NC
388taken (arguably for the wrong thing now, but on SunOS 4.1 it is the command
389for creating dynamically-loadable modules) and C<$link> could be confused with
390the Unix command line executable of the same name, which does something
391completely different. Andy Dougherty makes the counter argument "In parrot, I
392tried to call the command used to link object files and libraries into an
393executable F<link>, since that's what my vaguely-remembered DOS and VMS
394experience suggested. I don't think any real confusion has ensued, so it's
395probably a reasonable name for perl5 to use."
98fca0e8
NC
396
397"Alas, I've always worried that introducing it would make things worse,
398since now the module building utilities would have to look for
399C<$Config{link}> and institute a fall-back plan if it weren't found."
8d159ec1
NC
400Although I can see that as confusing, given that C<$Config{d_link}> is true
401when (hard) links are available.
98fca0e8 402
75585ce3
SP
403=head2 Configure Windows using PowerShell
404
405Currently, Windows uses hard-coded config files based to build the
406config.h for compiling Perl. Makefiles are also hard-coded and need to be
407hand edited prior to building Perl. While this makes it easy to create a perl.exe
408that works across multiple Windows versions, being able to accurately
409configure a perl.exe for a specific Windows versions and VS C++ would be
410a nice enhancement. With PowerShell available on Windows XP and up, this
411may now be possible. Step 1 might be to investigate whether this is possible
412and use this to clean up our current makefile situation. Step 2 would be to
413see if there would be a way to use our existing metaconfig units to configure a
414Windows Perl or whether we go in a separate direction and make it so. Of
415course, we all know what step 3 is.
416
ab45a0fa
NC
417=head2 decouple -g and -DDEBUGGING
418
419Currently F<Configure> automatically adds C<-DDEBUGGING> to the C compiler
420flags if it spots C<-g> in the optimiser flags. The pre-processor directive
eeab323f 421C<DEBUGGING> enables F<perl>'s command line C<-D> options, but in the process
ab45a0fa
NC
422makes F<perl> slower. It would be good to disentangle this logic, so that
423C-level debugging with C<-g> and Perl level debugging with C<-D> can easily
424be enabled independently.
425
0bdfc961
NC
426=head1 Tasks that need a little C knowledge
427
428These tasks would need a little C knowledge, but don't need any specific
429background or experience with XS, or how the Perl interpreter works
430
3d826b29
NC
431=head2 Weed out needless PERL_UNUSED_ARG
432
433The C code uses the macro C<PERL_UNUSED_ARG> to stop compilers warning about
434unused arguments. Often the arguments can't be removed, as there is an
435external constraint that determines the prototype of the function, so this
436approach is valid. However, there are some cases where C<PERL_UNUSED_ARG>
437could be removed. Specifically
438
439=over 4
440
441=item *
442
443The prototypes of (nearly all) static functions can be changed
444
445=item *
446
447Unused arguments generated by short cut macros are wasteful - the short cut
448macro used can be changed.
449
450=back
451
fbf638cb
RGS
452=head2 Modernize the order of directories in @INC
453
454The way @INC is laid out by default, one cannot upgrade core (dual-life)
455modules without overwriting files. This causes problems for binary
3d14fd97
AD
456package builders. One possible proposal is laid out in this
457message:
458L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2002-04/msg02380.html>.
fbf638cb 459
bcbaa2d5
RGS
460=head2 -Duse32bit*
461
462Natively 64-bit systems need neither -Duse64bitint nor -Duse64bitall.
463On these systems, it might be the default compilation mode, and there
464is currently no guarantee that passing no use64bitall option to the
465Configure process will build a 32bit perl. Implementing -Duse32bit*
fd2dadea 466options would be nice for perl 5.14.
bcbaa2d5 467
fee0a0f7 468=head2 Profile Perl - am I hot or not?
62403a3c 469
fee0a0f7
NC
470The Perl source code is stable enough that it makes sense to profile it,
471identify and optimise the hotspots. It would be good to measure the
472performance of the Perl interpreter using free tools such as cachegrind,
473gprof, and dtrace, and work to reduce the bottlenecks they reveal.
474
475As part of this, the idea of F<pp_hot.c> is that it contains the I<hot> ops,
476the ops that are most commonly used. The idea is that by grouping them, their
477object code will be adjacent in the executable, so they have a greater chance
478of already being in the CPU cache (or swapped in) due to being near another op
479already in use.
62403a3c
NC
480
481Except that it's not clear if these really are the most commonly used ops. So
fee0a0f7
NC
482as part of exercising your skills with coverage and profiling tools you might
483want to determine what ops I<really> are the most commonly used. And in turn
484suggest evictions and promotions to achieve a better F<pp_hot.c>.
62403a3c 485
91d0cbf6
NC
486One piece of Perl code that might make a good testbed is F<installman>.
487
98fed0ad
NC
488=head2 Allocate OPs from arenas
489
490Currently all new OP structures are individually malloc()ed and free()d.
491All C<malloc> implementations have space overheads, and are now as fast as
492custom allocates so it would both use less memory and less CPU to allocate
493the various OP structures from arenas. The SV arena code can probably be
494re-used for this.
495
539f2c54
JC
496Note that Configuring perl with C<-Accflags=-DPL_OP_SLAB_ALLOC> will use
497Perl_Slab_alloc() to pack optrees into a contiguous block, which is
498probably superior to the use of OP arenas, esp. from a cache locality
499standpoint. See L<Profile Perl - am I hot or not?>.
500
a229ae3b 501=head2 Improve win32/wince.c
0bdfc961 502
a229ae3b 503Currently, numerous functions look virtually, if not completely,
02f21748 504identical in both C<win32/wince.c> and C<win32/win32.c> files, which can't
6d71adcd
NC
505be good.
506
c5b31784
SH
507=head2 Use secure CRT functions when building with VC8 on Win32
508
509Visual C++ 2005 (VC++ 8.x) deprecated a number of CRT functions on the basis
510that they were "unsafe" and introduced differently named secure versions of
511them as replacements, e.g. instead of writing
512
513 FILE* f = fopen(__FILE__, "r");
514
515one should now write
516
517 FILE* f;
518 errno_t err = fopen_s(&f, __FILE__, "r");
519
520Currently, the warnings about these deprecations have been disabled by adding
521-D_CRT_SECURE_NO_DEPRECATE to the CFLAGS. It would be nice to remove that
522warning suppressant and actually make use of the new secure CRT functions.
523
524There is also a similar issue with POSIX CRT function names like fileno having
525been deprecated in favour of ISO C++ conformant names like _fileno. These
26a6faa8 526warnings are also currently suppressed by adding -D_CRT_NONSTDC_NO_DEPRECATE. It
c5b31784
SH
527might be nice to do as Microsoft suggest here too, although, unlike the secure
528functions issue, there is presumably little or no benefit in this case.
529
038ae9a4
SH
530=head2 Fix POSIX::access() and chdir() on Win32
531
532These functions currently take no account of DACLs and therefore do not behave
533correctly in situations where access is restricted by DACLs (as opposed to the
534read-only attribute).
535
536Furthermore, POSIX::access() behaves differently for directories having the
537read-only attribute set depending on what CRT library is being used. For
538example, the _access() function in the VC6 and VC7 CRTs (wrongly) claim that
539such directories are not writable, whereas in fact all directories are writable
540unless access is denied by DACLs. (In the case of directories, the read-only
541attribute actually only means that the directory cannot be deleted.) This CRT
542bug is fixed in the VC8 and VC9 CRTs (but, of course, the directory may still
543not actually be writable if access is indeed denied by DACLs).
544
545For the chdir() issue, see ActiveState bug #74552:
546http://bugs.activestate.com/show_bug.cgi?id=74552
547
548Therefore, DACLs should be checked both for consistency across CRTs and for
549the correct answer.
550
551(Note that perl's -w operator should not be modified to check DACLs. It has
552been written so that it reflects the state of the read-only attribute, even
553for directories (whatever CRT is being used), for symmetry with chmod().)
554
16815324
NC
555=head2 strcat(), strcpy(), strncat(), strncpy(), sprintf(), vsprintf()
556
557Maybe create a utility that checks after each libperl.a creation that
558none of the above (nor sprintf(), vsprintf(), or *SHUDDER* gets())
559ever creep back to libperl.a.
560
561 nm libperl.a | ./miniperl -alne '$o = $F[0] if /:$/; print "$o $F[1]" if $F[0] eq "U" && $F[1] =~ /^(?:strn?c(?:at|py)|v?sprintf|gets)$/'
562
563Note, of course, that this will only tell whether B<your> platform
564is using those naughty interfaces.
565
de96509d
JH
566=head2 -D_FORTIFY_SOURCE=2, -fstack-protector
567
568Recent glibcs support C<-D_FORTIFY_SOURCE=2> and recent gcc
569(4.1 onwards?) supports C<-fstack-protector>, both of which give
570protection against various kinds of buffer overflow problems.
571These should probably be used for compiling Perl whenever available,
572Configure and/or hints files should be adjusted to probe for the
573availability of these features and enable them as appropriate.
16815324 574
8964cfe0
NC
575=head2 Arenas for GPs? For MAGIC?
576
577C<struct gp> and C<struct magic> are both currently allocated by C<malloc>.
578It might be a speed or memory saving to change to using arenas. Or it might
579not. It would need some suitable benchmarking first. In particular, C<GP>s
580can probably be changed with minimal compatibility impact (probably nothing
581outside of the core, or even outside of F<gv.c> allocates them), but they
582probably aren't allocated/deallocated often enough for a speed saving. Whereas
583C<MAGIC> is allocated/deallocated more often, but in turn, is also something
584more externally visible, so changing the rules here may bite external code.
585
3880c8ec
NC
586=head2 Shared arenas
587
588Several SV body structs are now the same size, notably PVMG and PVGV, PVAV and
589PVHV, and PVCV and PVFM. It should be possible to allocate and return same
590sized bodies from the same actual arena, rather than maintaining one arena for
591each. This could save 4-6K per thread, of memory no longer tied up in the
592not-yet-allocated part of an arena.
593
8964cfe0 594
6d71adcd
NC
595=head1 Tasks that need a knowledge of XS
596
597These tasks would need C knowledge, and roughly the level of knowledge of
598the perl API that comes from writing modules that use XS to interface to
599C.
600
e851c105
DG
601=head2 Write an XS cookbook
602
603Create pod/perlxscookbook.pod with short, task-focused 'recipes' in XS that
604demonstrate common tasks and good practices. (Some of these might be
605extracted from perlguts.) The target audience should be XS novices, who need
606more examples than perlguts but something less overwhelming than perlapi.
607Recipes should provide "one pretty good way to do it" instead of TIMTOWTDI.
608
5b7d14ff
DG
609Rather than focusing on interfacing Perl to C libraries, such a cookbook
610should probably focus on how to optimize Perl routines by re-writing them
611in XS. This will likely be more motivating to those who mostly work in
612Perl but are looking to take the next step into XS.
613
614Deconstructing and explaining some simpler XS modules could be one way to
615bootstrap a cookbook. (List::Util? Class::XSAccessor? Tree::Ternary_XS?)
616Another option could be deconstructing the implementation of some simpler
617functions in op.c.
618
05fb4e20
NC
619=head2 Allow XSUBs to inline themselves as OPs
620
621For a simple XSUB, often the subroutine dispatch takes more time than the
622XSUB itself. The tokeniser already has the ability to inline constant
623subroutines - it would be good to provide a way to inline other subroutines.
624
625Specifically, simplest approach looks to be to allow an XSUB to provide an
626alternative implementation of itself as a custom OP. A new flag bit in
627C<CvFLAGS()> would signal to the peephole optimiser to take an optree
628such as this:
629
630 b <@> leave[1 ref] vKP/REFC ->(end)
631 1 <0> enter ->2
632 2 <;> nextstate(main 1 -e:1) v:{ ->3
633 a <2> sassign vKS/2 ->b
634 8 <1> entersub[t2] sKS/TARG,1 ->9
635 - <1> ex-list sK ->8
636 3 <0> pushmark s ->4
637 4 <$> const(IV 1) sM ->5
638 6 <1> rv2av[t1] lKM/1 ->7
639 5 <$> gv(*a) s ->6
640 - <1> ex-rv2cv sK ->-
641 7 <$> gv(*x) s/EARLYCV ->8
642 - <1> ex-rv2sv sKRM*/1 ->a
643 9 <$> gvsv(*b) s ->a
644
645perform the symbol table lookup of C<rv2cv> and C<gv(*x)>, locate the
646pointer to the custom OP that provides the direct implementation, and re-
647write the optree something like:
648
649 b <@> leave[1 ref] vKP/REFC ->(end)
650 1 <0> enter ->2
651 2 <;> nextstate(main 1 -e:1) v:{ ->3
652 a <2> sassign vKS/2 ->b
653 7 <1> custom_x -> 8
654 - <1> ex-list sK ->7
655 3 <0> pushmark s ->4
656 4 <$> const(IV 1) sM ->5
657 6 <1> rv2av[t1] lKM/1 ->7
658 5 <$> gv(*a) s ->6
659 - <1> ex-rv2cv sK ->-
660 - <$> ex-gv(*x) s/EARLYCV ->7
661 - <1> ex-rv2sv sKRM*/1 ->a
662 8 <$> gvsv(*b) s ->a
663
664I<i.e.> the C<gv(*)> OP has been nulled and spliced out of the execution
665path, and the C<entersub> OP has been replaced by the custom op.
666
667This approach should provide a measurable speed up to simple XSUBs inside
668tight loops. Initially one would have to write the OP alternative
669implementation by hand, but it's likely that this should be reasonably
670straightforward for the type of XSUB that would benefit the most. Longer
671term, once the run-time implementation is proven, it should be possible to
672progressively update ExtUtils::ParseXS to generate OP implementations for
673some XSUBs.
674
318bf708
NC
675=head2 Remove the use of SVs as temporaries in dump.c
676
677F<dump.c> contains debugging routines to dump out the contains of perl data
678structures, such as C<SV>s, C<AV>s and C<HV>s. Currently, the dumping code
679B<uses> C<SV>s for its temporary buffers, which was a logical initial
680implementation choice, as they provide ready made memory handling.
681
682However, they also lead to a lot of confusion when it happens that what you're
683trying to debug is seen by the code in F<dump.c>, correctly or incorrectly, as
684a temporary scalar it can use for a temporary buffer. It's also not possible
685to dump scalars before the interpreter is properly set up, such as during
686ithreads cloning. It would be good to progressively replace the use of scalars
687as string accumulation buffers with something much simpler, directly allocated
688by C<malloc>. The F<dump.c> code is (or should be) only producing 7 bit
689US-ASCII, so output character sets are not an issue.
690
691Producing and proving an internal simple buffer allocation would make it easier
692to re-write the internals of the PerlIO subsystem to avoid using C<SV>s for
693B<its> buffers, use of which can cause problems similar to those of F<dump.c>,
694at similar times.
695
5d96f598
NC
696=head2 safely supporting POSIX SA_SIGINFO
697
698Some years ago Jarkko supplied patches to provide support for the POSIX
699SA_SIGINFO feature in Perl, passing the extra data to the Perl signal handler.
700
701Unfortunately, it only works with "unsafe" signals, because under safe
702signals, by the time Perl gets to run the signal handler, the extra
703information has been lost. Moreover, it's not easy to store it somewhere,
704as you can't call mutexs, or do anything else fancy, from inside a signal
705handler.
706
707So it strikes me that we could provide safe SA_SIGINFO support
708
709=over 4
710
711=item 1
712
713Provide global variables for two file descriptors
714
715=item 2
716
717When the first request is made via C<sigaction> for C<SA_SIGINFO>, create a
718pipe, store the reader in one, the writer in the other
719
720=item 3
721
722In the "safe" signal handler (C<Perl_csighandler()>/C<S_raise_signal()>), if
723the C<siginfo_t> pointer non-C<NULL>, and the writer file handle is open,
724
725=over 8
726
727=item 1
728
729serialise signal number, C<struct siginfo_t> (or at least the parts we care
730about) into a small auto char buff
731
732=item 2
733
734C<write()> that (non-blocking) to the writer fd
735
736=over 12
737
738=item 1
739
740if it writes 100%, flag the signal in a counter of "signals on the pipe" akin
741to the current per-signal-number counts
742
743=item 2
744
745if it writes 0%, assume the pipe is full. Flag the data as lost?
746
747=item 3
748
749if it writes partially, croak a panic, as your OS is broken.
750
751=back
752
753=back
754
755=item 4
756
757in the regular C<PERL_ASYNC_CHECK()> processing, if there are "signals on
758the pipe", read the data out, deserialise, build the Perl structures on
759the stack (code in C<Perl_sighandler()>, the "unsafe" handler), and call as
760usual.
761
762=back
763
764I think that this gets us decent C<SA_SIGINFO> support, without the current risk
765of running Perl code inside the signal handler context. (With all the dangers
766of things like C<malloc> corruption that that currently offers us)
767
768For more information see the thread starting with this message:
769http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-03/msg00305.html
770
6d71adcd
NC
771=head2 autovivification
772
773Make all autovivification consistent w.r.t LVALUE/RVALUE and strict/no strict;
774
775This task is incremental - even a little bit of work on it will help.
776
777=head2 Unicode in Filenames
778
779chdir, chmod, chown, chroot, exec, glob, link, lstat, mkdir, open,
780opendir, qx, readdir, readlink, rename, rmdir, stat, symlink, sysopen,
781system, truncate, unlink, utime, -X. All these could potentially accept
782Unicode filenames either as input or output (and in the case of system
783and qx Unicode in general, as input or output to/from the shell).
784Whether a filesystem - an operating system pair understands Unicode in
785filenames varies.
786
787Known combinations that have some level of understanding include
788Microsoft NTFS, Apple HFS+ (In Mac OS 9 and X) and Apple UFS (in Mac
789OS X), NFS v4 is rumored to be Unicode, and of course Plan 9. How to
790create Unicode filenames, what forms of Unicode are accepted and used
791(UCS-2, UTF-16, UTF-8), what (if any) is the normalization form used,
792and so on, varies. Finding the right level of interfacing to Perl
793requires some thought. Remember that an OS does not implicate a
794filesystem.
795
796(The Windows -C command flag "wide API support" has been at least
797temporarily retired in 5.8.1, and the -C has been repurposed, see
798L<perlrun>.)
799
87a942b1
JH
800Most probably the right way to do this would be this:
801L</"Virtualize operating system access">.
802
6d71adcd
NC
803=head2 Unicode in %ENV
804
805Currently the %ENV entries are always byte strings.
87a942b1 806See L</"Virtualize operating system access">.
6d71adcd 807
1f2e7916
JD
808=head2 Unicode and glob()
809
810Currently glob patterns and filenames returned from File::Glob::glob()
87a942b1 811are always byte strings. See L</"Virtualize operating system access">.
1f2e7916 812
6d71adcd
NC
813=head2 use less 'memory'
814
815Investigate trade offs to switch out perl's choices on memory usage.
816Particularly perl should be able to give memory back.
817
818This task is incremental - even a little bit of work on it will help.
819
820=head2 Re-implement C<:unique> in a way that is actually thread-safe
821
822The old implementation made bad assumptions on several levels. A good 90%
823solution might be just to make C<:unique> work to share the string buffer
824of SvPVs. That way large constant strings can be shared between ithreads,
825such as the configuration information in F<Config>.
826
827=head2 Make tainting consistent
828
829Tainting would be easier to use if it didn't take documented shortcuts and
830allow taint to "leak" everywhere within an expression.
831
832=head2 readpipe(LIST)
833
834system() accepts a LIST syntax (and a PROGRAM LIST syntax) to avoid
835running a shell. readpipe() (the function behind qx//) could be similarly
836extended.
837
6d71adcd
NC
838=head2 Audit the code for destruction ordering assumptions
839
840Change 25773 notes
841
842 /* Need to check SvMAGICAL, as during global destruction it may be that
843 AvARYLEN(av) has been freed before av, and hence the SvANY() pointer
844 is now part of the linked list of SV heads, rather than pointing to
845 the original body. */
846 /* FIXME - audit the code for other bugs like this one. */
847
848adding the C<SvMAGICAL> check to
849
850 if (AvARYLEN(av) && SvMAGICAL(AvARYLEN(av))) {
851 MAGIC *mg = mg_find (AvARYLEN(av), PERL_MAGIC_arylen);
852
853Go through the core and look for similar assumptions that SVs have particular
854types, as all bets are off during global destruction.
855
749904bf
JH
856=head2 Extend PerlIO and PerlIO::Scalar
857
858PerlIO::Scalar doesn't know how to truncate(). Implementing this
859would require extending the PerlIO vtable.
860
861Similarly the PerlIO vtable doesn't know about formats (write()), or
862about stat(), or chmod()/chown(), utime(), or flock().
863
864(For PerlIO::Scalar it's hard to see what e.g. mode bits or ownership
865would mean.)
866
867PerlIO doesn't do directories or symlinks, either: mkdir(), rmdir(),
868opendir(), closedir(), seekdir(), rewinddir(), glob(); symlink(),
869readlink().
870
94da6c29
JH
871See also L</"Virtualize operating system access">.
872
3236f110
NC
873=head2 -C on the #! line
874
875It should be possible to make -C work correctly if found on the #! line,
876given that all perl command line options are strict ASCII, and -C changes
877only the interpretation of non-ASCII characters, and not for the script file
878handle. To make it work needs some investigation of the ordering of function
879calls during startup, and (by implication) a bit of tweaking of that order.
880
d6c1e11f
JH
881=head2 Organize error messages
882
883Perl's diagnostics (error messages, see L<perldiag>) could use
a8d0aeb9 884reorganizing and formalizing so that each error message has its
d6c1e11f
JH
885stable-for-all-eternity unique id, categorized by severity, type, and
886subsystem. (The error messages would be listed in a datafile outside
c4bd451b
CB
887of the Perl source code, and the source code would only refer to the
888messages by the id.) This clean-up and regularizing should apply
d6c1e11f
JH
889for all croak() messages.
890
891This would enable all sorts of things: easier translation/localization
892of the messages (though please do keep in mind the caveats of
893L<Locale::Maketext> about too straightforward approaches to
894translation), filtering by severity, and instead of grepping for a
895particular error message one could look for a stable error id. (Of
896course, changing the error messages by default would break all the
897existing software depending on some particular error message...)
898
899This kind of functionality is known as I<message catalogs>. Look for
900inspiration for example in the catgets() system, possibly even use it
901if available-- but B<only> if available, all platforms will B<not>
de96509d 902have catgets().
d6c1e11f
JH
903
904For the really pure at heart, consider extending this item to cover
905also the warning messages (see L<perllexwarn>, C<warnings.pl>).
3236f110 906
0bdfc961 907=head1 Tasks that need a knowledge of the interpreter
3298bd4d 908
0bdfc961
NC
909These tasks would need C knowledge, and knowledge of how the interpreter works,
910or a willingness to learn.
3298bd4d 911
10517af5
JD
912=head2 forbid labels with keyword names
913
914Currently C<goto keyword> "computes" the label value:
915
916 $ perl -e 'goto print'
917 Can't find label 1 at -e line 1.
918
343c8006
JD
919It is controversial if the right way to avoid the confusion is to forbid
920labels with keyword names, or if it would be better to always treat
921bareword expressions after a "goto" as a label and never as a keyword.
10517af5 922
de6375e3
RGS
923=head2 truncate() prototype
924
925The prototype of truncate() is currently C<$$>. It should probably
926be C<*$> instead. (This is changed in F<opcode.pl>)
927
2d0587d8
RGS
928=head2 decapsulation of smart match argument
929
930Currently C<$foo ~~ $object> will die with the message "Smart matching a
931non-overloaded object breaks encapsulation". It would be nice to allow
932to bypass this by using explictly the syntax C<$foo ~~ %$object> or
933C<$foo ~~ @$object>.
934
565590b5
NC
935=head2 error reporting of [$a ; $b]
936
937Using C<;> inside brackets is a syntax error, and we don't propose to change
938that by giving it any meaning. However, it's not reported very helpfully:
939
940 $ perl -e '$a = [$b; $c];'
941 syntax error at -e line 1, near "$b;"
942 syntax error at -e line 1, near "$c]"
943 Execution of -e aborted due to compilation errors.
944
945It should be possible to hook into the tokeniser or the lexer, so that when a
946C<;> is parsed where it is not legal as a statement terminator (ie inside
947C<{}> used as a hashref, C<[]> or C<()>) it issues an error something like
948I<';' isn't legal inside an expression - if you need multiple statements use a
949do {...} block>. See the thread starting at
950http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-09/msg00573.html
951
718140ec
NC
952=head2 lexicals used only once
953
954This warns:
955
956 $ perl -we '$pie = 42'
957 Name "main::pie" used only once: possible typo at -e line 1.
958
959This does not:
960
961 $ perl -we 'my $pie = 42'
962
963Logically all lexicals used only once should warn, if the user asks for
d6f4ea2e
SP
964warnings. An unworked RT ticket (#5087) has been open for almost seven
965years for this discrepancy.
718140ec 966
a3d15f9a
RGS
967=head2 UTF-8 revamp
968
85c006b6
KW
969The handling of Unicode is unclean in many places. In the regex engine
970there are especially many problems. The swash data structure could be
971replaced my something better. Inversion lists and maps are likely
972candidates. The whole Unicode database could be placed in-core for a
973huge speed-up. Only minimal work was done on the optimizer when utf8
974was added, with the result that the synthetic start class often will
975fail to narrow down the possible choices when given non-Latin1 input.
a3d15f9a
RGS
976
977=head2 Properly Unicode safe tokeniser and pads.
978
979The tokeniser isn't actually very UTF-8 clean. C<use utf8;> is a hack -
980variable names are stored in stashes as raw bytes, without the utf-8 flag
981set. The pad API only takes a C<char *> pointer, so that's all bytes too. The
982tokeniser ignores the UTF-8-ness of C<PL_rsfp>, or any SVs returned from
983source filters. All this could be fixed.
984
636e63cb
NC
985=head2 state variable initialization in list context
986
987Currently this is illegal:
988
989 state ($a, $b) = foo();
990
a2874905 991In Perl 6, C<state ($a) = foo();> and C<(state $a) = foo();> have different
a8d0aeb9 992semantics, which is tricky to implement in Perl 5 as currently they produce
a2874905 993the same opcode trees. The Perl 6 design is firm, so it would be good to
a8d0aeb9 994implement the necessary code in Perl 5. There are comments in
a2874905
NC
995C<Perl_newASSIGNOP()> that show the code paths taken by various assignment
996constructions involving state variables.
636e63cb 997
4fedb12c
RGS
998=head2 Implement $value ~~ 0 .. $range
999
1000It would be nice to extend the syntax of the C<~~> operator to also
1001understand numeric (and maybe alphanumeric) ranges.
a393eb28
RGS
1002
1003=head2 A does() built-in
1004
1005Like ref(), only useful. It would call the C<DOES> method on objects; it
1006would also tell whether something can be dereferenced as an
1007array/hash/etc., or used as a regexp, etc.
1008L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-03/msg00481.html>
1009
1010=head2 Tied filehandles and write() don't mix
1011
1012There is no method on tied filehandles to allow them to be called back by
1013formats.
4fedb12c 1014
53967bb9
RGS
1015=head2 Propagate compilation hints to the debugger
1016
1017Currently a debugger started with -dE on the command-line doesn't see the
1018features enabled by -E. More generally hints (C<$^H> and C<%^H>) aren't
1019propagated to the debugger. Probably it would be a good thing to propagate
1020hints from the innermost non-C<DB::> scope: this would make code eval'ed
1021in the debugger see the features (and strictures, etc.) currently in
1022scope.
1023
d10fc472 1024=head2 Attach/detach debugger from running program
1626a787 1025
cd793d32
NC
1026The old perltodo notes "With C<gdb>, you can attach the debugger to a running
1027program if you pass the process ID. It would be good to do this with the Perl
0bdfc961
NC
1028debugger on a running Perl program, although I'm not sure how it would be
1029done." ssh and screen do this with named pipes in /tmp. Maybe we can too.
1626a787 1030
0bdfc961
NC
1031=head2 LVALUE functions for lists
1032
1033The old perltodo notes that lvalue functions don't work for list or hash
1034slices. This would be good to fix.
1035
0bdfc961
NC
1036=head2 regexp optimiser optional
1037
1038The regexp optimiser is not optional. It should configurable to be, to allow
1039its performance to be measured, and its bugs to be easily demonstrated.
1040
ef36c6a7
RGS
1041=head2 C</w> regex modifier
1042
1043That flag would enable to match whole words, and also to interpolate
1044arrays as alternations. With it, C</P/w> would be roughly equivalent to:
1045
1046 do { local $"='|'; /\b(?:P)\b/ }
1047
1048See L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-01/msg00400.html>
1049for the discussion.
1050
0bdfc961
NC
1051=head2 optional optimizer
1052
1053Make the peephole optimizer optional. Currently it performs two tasks as
1054it walks the optree - genuine peephole optimisations, and necessary fixups of
1055ops. It would be good to find an efficient way to switch out the
1056optimisations whilst keeping the fixups.
1057
1058=head2 You WANT *how* many
1059
1060Currently contexts are void, scalar and list. split has a special mechanism in
1061place to pass in the number of return values wanted. It would be useful to
1062have a general mechanism for this, backwards compatible and little speed hit.
1063This would allow proposals such as short circuiting sort to be implemented
1064as a module on CPAN.
1065
1066=head2 lexical aliases
1067
1068Allow lexical aliases (maybe via the syntax C<my \$alias = \$foo>.
1069
1070=head2 entersub XS vs Perl
1071
1072At the moment pp_entersub is huge, and has code to deal with entering both
1073perl and XS subroutines. Subroutine implementations rarely change between
1074perl and XS at run time, so investigate using 2 ops to enter subs (one for
1075XS, one for perl) and swap between if a sub is redefined.
2810d901 1076
de535794 1077=head2 Self-ties
2810d901 1078
de535794 1079Self-ties are currently illegal because they caused too many segfaults. Maybe
a8d0aeb9 1080the causes of these could be tracked down and self-ties on all types
de535794 1081reinstated.
0bdfc961
NC
1082
1083=head2 Optimize away @_
1084
1085The old perltodo notes "Look at the "reification" code in C<av.c>".
1086
87a942b1
JH
1087=head2 Virtualize operating system access
1088
1089Implement a set of "vtables" that virtualizes operating system access
1090(open(), mkdir(), unlink(), readdir(), getenv(), etc.) At the very
1091least these interfaces should take SVs as "name" arguments instead of
1092bare char pointers; probably the most flexible and extensible way
e1a3d5d1
JH
1093would be for the Perl-facing interfaces to accept HVs. The system
1094needs to be per-operating-system and per-file-system
1095hookable/filterable, preferably both from XS and Perl level
87a942b1
JH
1096(L<perlport/"Files and Filesystems"> is good reading at this point,
1097in fact, all of L<perlport> is.)
1098
e1a3d5d1
JH
1099This has actually already been implemented (but only for Win32),
1100take a look at F<iperlsys.h> and F<win32/perlhost.h>. While all Win32
1101variants go through a set of "vtables" for operating system access,
e1020413 1102non-Win32 systems currently go straight for the POSIX/Unix-style
e1a3d5d1
JH
1103system/library call. Similar system as for Win32 should be
1104implemented for all platforms. The existing Win32 implementation
1105probably does not need to survive alongside this proposed new
1106implementation, the approaches could be merged.
87a942b1
JH
1107
1108What would this give us? One often-asked-for feature this would
94da6c29
JH
1109enable is using Unicode for filenames, and other "names" like %ENV,
1110usernames, hostnames, and so forth.
1111(See L<perlunicode/"When Unicode Does Not Happen">.)
1112
1113But this kind of virtualization would also allow for things like
1114virtual filesystems, virtual networks, and "sandboxes" (though as long
1115as dynamic loading of random object code is allowed, not very safe
1116sandboxes since external code of course know not of Perl's vtables).
1117An example of a smaller "sandbox" is that this feature can be used to
1118implement per-thread working directories: Win32 already does this.
1119
1120See also L</"Extend PerlIO and PerlIO::Scalar">.
87a942b1 1121
ac6197af
NC
1122=head2 Investigate PADTMP hash pessimisation
1123
9a2f2e6b 1124The peephole optimiser converts constants used for hash key lookups to shared
057163d7 1125hash key scalars. Under ithreads, something is undoing this work.
ac6197af
NC
1126See http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-09/msg00793.html
1127
057163d7
NC
1128=head2 Store the current pad in the OP slab allocator
1129
1130=for clarification
1131I hope that I got that "current pad" part correct
1132
1133Currently we leak ops in various cases of parse failure. I suggested that we
1134could solve this by always using the op slab allocator, and walking it to
1135free ops. Dave comments that as some ops are already freed during optree
1136creation one would have to mark which ops are freed, and not double free them
1137when walking the slab. He notes that one problem with this is that for some ops
1138you have to know which pad was current at the time of allocation, which does
1139change. I suggested storing a pointer to the current pad in the memory allocated
1140for the slab, and swapping to a new slab each time the pad changes. Dave thinks
1141that this would work.
1142
52960e22
JC
1143=head2 repack the optree
1144
1145Repacking the optree after execution order is determined could allow
057163d7
NC
1146removal of NULL ops, and optimal ordering of OPs with respect to cache-line
1147filling. The slab allocator could be reused for this purpose. I think that
1148the best way to do this is to make it an optional step just before the
1149completed optree is attached to anything else, and to use the slab allocator
1150unchanged, so that freeing ops is identical whether or not this step runs.
1151Note that the slab allocator allocates ops downwards in memory, so one would
1152have to actually "allocate" the ops in reverse-execution order to get them
1153contiguous in memory in execution order.
1154
1155See http://www.nntp.perl.org/group/perl.perl5.porters/2007/12/msg131975.html
1156
1157Note that running this copy, and then freeing all the old location ops would
1158cause their slabs to be freed, which would eliminate possible memory wastage if
1159the previous suggestion is implemented, and we swap slabs more frequently.
52960e22 1160
12e06b6f
NC
1161=head2 eliminate incorrect line numbers in warnings
1162
1163This code
1164
1165 use warnings;
1166 my $undef;
1167
1168 if ($undef == 3) {
1169 } elsif ($undef == 0) {
1170 }
1171
18a16cc5 1172used to produce this output:
12e06b6f
NC
1173
1174 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1175 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1176
18a16cc5
NC
1177where the line of the second warning was misreported - it should be line 5.
1178Rafael fixed this - the problem arose because there was no nextstate OP
1179between the execution of the C<if> and the C<elsif>, hence C<PL_curcop> still
1180reports that the currently executing line is line 4. The solution was to inject
1181a nextstate OPs for each C<elsif>, although it turned out that the nextstate
1182OP needed to be a nulled OP, rather than a live nextstate OP, else other line
1183numbers became misreported. (Jenga!)
12e06b6f
NC
1184
1185The problem is more general than C<elsif> (although the C<elsif> case is the
1186most common and the most confusing). Ideally this code
1187
1188 use warnings;
1189 my $undef;
1190
1191 my $a = $undef + 1;
1192 my $b
1193 = $undef
1194 + 1;
1195
1196would produce this output
1197
1198 Use of uninitialized value $undef in addition (+) at wrong.pl line 4.
1199 Use of uninitialized value $undef in addition (+) at wrong.pl line 7.
1200
1201(rather than lines 4 and 5), but this would seem to require every OP to carry
1202(at least) line number information.
1203
1204What might work is to have an optional line number in memory just before the
1205BASEOP structure, with a flag bit in the op to say whether it's present.
1206Initially during compile every OP would carry its line number. Then add a late
1207pass to the optimiser (potentially combined with L</repack the optree>) which
1208looks at the two ops on every edge of the graph of the execution path. If
1209the line number changes, flags the destination OP with this information.
1210Once all paths are traced, replace every op with the flag with a
1211nextstate-light op (that just updates C<PL_curcop>), which in turn then passes
1212control on to the true op. All ops would then be replaced by variants that
1213do not store the line number. (Which, logically, why it would work best in
1214conjunction with L</repack the optree>, as that is already copying/reallocating
1215all the OPs)
1216
18a16cc5
NC
1217(Although I should note that we're not certain that doing this for the general
1218case is worth it)
1219
52960e22
JC
1220=head2 optimize tail-calls
1221
1222Tail-calls present an opportunity for broadly applicable optimization;
1223anywhere that C<< return foo(...) >> is called, the outer return can
1224be replaced by a goto, and foo will return directly to the outer
1225caller, saving (conservatively) 25% of perl's call&return cost, which
1226is relatively higher than in C. The scheme language is known to do
1227this heavily. B::Concise provides good insight into where this
1228optimization is possible, ie anywhere entersub,leavesub op-sequence
1229occurs.
1230
1231 perl -MO=Concise,-exec,a,b,-main -e 'sub a{ 1 }; sub b {a()}; b(2)'
1232
1233Bottom line on this is probably a new pp_tailcall function which
1234combines the code in pp_entersub, pp_leavesub. This should probably
1235be done 1st in XS, and using B::Generate to patch the new OP into the
1236optrees.
1237
0c397127
KW
1238=head2 Add C<00dddd>
1239
1240It has been proposed that octal constants be specifiable through the syntax
1241C<0oddddd>, parallel to the existing construct to specify hex constants
1242C<0xddddd>
1243
0bdfc961
NC
1244=head1 Big projects
1245
1246Tasks that will get your name mentioned in the description of the "Highlights
fd2dadea 1247of 5.14"
0bdfc961
NC
1248
1249=head2 make ithreads more robust
1250
4e577f8b 1251Generally make ithreads more robust. See also L</iCOW>
0bdfc961
NC
1252
1253This task is incremental - even a little bit of work on it will help, and
1254will be greatly appreciated.
1255
07577ec1
FC
1256One bit would be to determine how to clone directory handles on systems
1257without a C<fchdir> function (in sv.c:Perl_dirp_dup).
6c047da7 1258
59c7f7d5
RGS
1259Fix Perl_sv_dup, et al so that threads can return objects.
1260
0bdfc961
NC
1261=head2 iCOW
1262
1263Sarathy and Arthur have a proposal for an improved Copy On Write which
1264specifically will be able to COW new ithreads. If this can be implemented
1265it would be a good thing.
1266
1267=head2 (?{...}) closures in regexps
1268
1269Fix (or rewrite) the implementation of the C</(?{...})/> closures.
1270
6bda09f9
YO
1271=head2 Add class set operations to regexp engine
1272
1273Apparently these are quite useful. Anyway, Jeffery Friedl wants them.
1274
1275demerphq has this on his todo list, but right at the bottom.
44a7a252
JV
1276
1277
1278=head1 Tasks for microperl
1279
1280
1281[ Each and every one of these may be obsolete, but they were listed
1282 in the old Todo.micro file]
1283
1284
1285=head2 make creating uconfig.sh automatic
1286
1287=head2 make creating Makefile.micro automatic
1288
1289=head2 do away with fork/exec/wait?
1290
1291(system, popen should be enough?)
1292
1293=head2 some of the uconfig.sh really needs to be probed (using cc) in buildtime:
1294
1295(uConfigure? :-) native datatype widths and endianness come to mind
1296