This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
perltodo.pod: Add more detail about @INC order.
[perl5.git] / pod / perltodo.pod
CommitLineData
7711098a
GS
1=head1 NAME
2
3perltodo - Perl TO-DO List
4
5=head1 DESCRIPTION
e50bb9a1 6
0bdfc961
NC
7This is a list of wishes for Perl. The tasks we think are smaller or easier
8are listed first. Anyone is welcome to work on any of these, but it's a good
9idea to first contact I<perl5-porters@perl.org> to avoid duplication of
10effort. By all means contact a pumpking privately first if you prefer.
e50bb9a1 11
0bdfc961
NC
12Whilst patches to make the list shorter are most welcome, ideas to add to
13the list are also encouraged. Check the perl5-porters archives for past
14ideas, and any discussion about them. One set of archives may be found at:
e50bb9a1 15
0bdfc961 16 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/
938c8732 17
617eabfa
NC
18What can we offer you in return? Fame, fortune, and everlasting glory? Maybe
19not, but if your patch is incorporated, then we'll add your name to the
20F<AUTHORS> file, which ships in the official distribution. How many other
21programming languages offer you 1 line of immortality?
938c8732 22
0bdfc961 23=head1 Tasks that only need Perl knowledge
e50bb9a1 24
0bdfc961 25=head2 common test code for timed bail out
e50bb9a1 26
0bdfc961
NC
27Write portable self destruct code for tests to stop them burning CPU in
28infinite loops. This needs to avoid using alarm, as some of the tests are
29testing alarm/sleep or timers.
e50bb9a1 30
87a942b1 31=head2 POD -E<gt> HTML conversion in the core still sucks
e50bb9a1 32
938c8732 33Which is crazy given just how simple POD purports to be, and how simple HTML
adebf063
NC
34can be. It's not actually I<as> simple as it sounds, particularly with the
35flexibility POD allows for C<=item>, but it would be good to improve the
36visual appeal of the HTML generated, and to avoid it having any validation
37errors. See also L</make HTML install work>, as the layout of installation tree
38is needed to improve the cross-linking.
938c8732 39
dc0fb092
SP
40The addition of C<Pod::Simple> and its related modules may make this task
41easier to complete.
42
aa237293
NC
43=head2 Parallel testing
44
b2e2905c 45(This probably impacts much more than the core: also the Test::Harness
02f21748
RGS
46and TAP::* modules on CPAN.)
47
aa237293
NC
48The core regression test suite is getting ever more comprehensive, which has
49the side effect that it takes longer to run. This isn't so good. Investigate
50whether it would be feasible to give the harness script the B<option> of
51running sets of tests in parallel. This would be useful for tests in
52F<t/op/*.t> and F<t/uni/*.t> and maybe some sets of tests in F<lib/>.
53
54Questions to answer
55
56=over 4
57
58=item 1
59
60How does screen layout work when you're running more than one test?
61
62=item 2
63
64How does the caller of test specify how many tests to run in parallel?
65
66=item 3
67
68How do setup/teardown tests identify themselves?
69
70=back
71
72Pugs already does parallel testing - can their approach be re-used?
73
0bdfc961 74=head2 Make Schwern poorer
e50bb9a1 75
613bd4f7 76We should have tests for everything. When all the core's modules are tested,
0bdfc961
NC
77Schwern has promised to donate to $500 to TPF. We may need volunteers to
78hold him upside down and shake vigorously in order to actually extract the
79cash.
3958b146 80
0bdfc961 81=head2 Improve the coverage of the core tests
e50bb9a1 82
02f21748
RGS
83Use Devel::Cover to ascertain the core modules's test coverage, then add
84tests that are currently missing.
30222c0f 85
0bdfc961 86=head2 test B
e50bb9a1 87
0bdfc961 88A full test suite for the B module would be nice.
e50bb9a1 89
636e63cb
NC
90=head2 Deparse inlined constants
91
92Code such as this
93
94 use constant PI => 4;
95 warn PI
96
97will currently deparse as
98
99 use constant ('PI', 4);
100 warn 4;
101
102because the tokenizer inlines the value of the constant subroutine C<PI>.
103This allows various compile time optimisations, such as constant folding
104and dead code elimination. Where these haven't happened (such as the example
105above) it ought be possible to make B::Deparse work out the name of the
106original constant, because just enough information survives in the symbol
107table to do this. Specifically, the same scalar is used for the constant in
108the optree as is used for the constant subroutine, so by iterating over all
109symbol tables and generating a mapping of SV address to constant name, it
110would be possible to provide B::Deparse with this functionality.
111
0bdfc961 112=head2 A decent benchmark
e50bb9a1 113
617eabfa 114C<perlbench> seems impervious to any recent changes made to the perl core. It
0bdfc961
NC
115would be useful to have a reasonable general benchmarking suite that roughly
116represented what current perl programs do, and measurably reported whether
117tweaks to the core improve, degrade or don't really affect performance, to
118guide people attempting to optimise the guts of perl. Gisle would welcome
119new tests for perlbench.
6168cf99 120
0bdfc961 121=head2 fix tainting bugs
6168cf99 122
0bdfc961
NC
123Fix the bugs revealed by running the test suite with the C<-t> switch (via
124C<make test.taintwarn>).
e50bb9a1 125
0bdfc961 126=head2 Dual life everything
e50bb9a1 127
0bdfc961
NC
128As part of the "dists" plan, anything that doesn't belong in the smallest perl
129distribution needs to be dual lifed. Anything else can be too. Figure out what
130changes would be needed to package that module and its tests up for CPAN, and
131do so. Test it with older perl releases, and fix the problems you find.
e50bb9a1 132
a393eb28
RGS
133To make a minimal perl distribution, it's useful to look at
134F<t/lib/commonsense.t>.
135
0bdfc961 136=head2 Improving C<threads::shared>
722d2a37 137
0bdfc961
NC
138Investigate whether C<threads::shared> could share aggregates properly with
139only Perl level changes to shared.pm
722d2a37 140
0bdfc961 141=head2 POSIX memory footprint
e50bb9a1 142
0bdfc961
NC
143Ilya observed that use POSIX; eats memory like there's no tomorrow, and at
144various times worked to cut it down. There is probably still fat to cut out -
145for example POSIX passes Exporter some very memory hungry data structures.
e50bb9a1 146
eed36644
NC
147=head2 embed.pl/makedef.pl
148
149There is a script F<embed.pl> that generates several header files to prefix
150all of Perl's symbols in a consistent way, to provide some semblance of
151namespace support in C<C>. Functions are declared in F<embed.fnc>, variables
907b3e23 152in F<interpvar.h>. Quite a few of the functions and variables
eed36644
NC
153are conditionally declared there, using C<#ifdef>. However, F<embed.pl>
154doesn't understand the C macros, so the rules about which symbols are present
155when is duplicated in F<makedef.pl>. Writing things twice is bad, m'kay.
156It would be good to teach C<embed.pl> to understand the conditional
157compilation, and hence remove the duplication, and the mistakes it has caused.
e50bb9a1 158
801de10e
NC
159=head2 use strict; and AutoLoad
160
161Currently if you write
162
163 package Whack;
164 use AutoLoader 'AUTOLOAD';
165 use strict;
166 1;
167 __END__
168 sub bloop {
169 print join (' ', No, strict, here), "!\n";
170 }
171
172then C<use strict;> isn't in force within the autoloaded subroutines. It would
173be more consistent (and less surprising) to arrange for all lexical pragmas
174in force at the __END__ block to be in force within each autoloaded subroutine.
175
773b3597
RGS
176There's a similar problem with SelfLoader.
177
0bdfc961 178=head1 Tasks that need a little sysadmin-type knowledge
e50bb9a1 179
0bdfc961
NC
180Or if you prefer, tasks that you would learn from, and broaden your skills
181base...
e50bb9a1 182
cd793d32 183=head2 make HTML install work
e50bb9a1 184
adebf063
NC
185There is an C<installhtml> target in the Makefile. It's marked as
186"experimental". It would be good to get this tested, make it work reliably, and
187remove the "experimental" tag. This would include
188
189=over 4
190
191=item 1
192
193Checking that cross linking between various parts of the documentation works.
194In particular that links work between the modules (files with POD in F<lib/>)
195and the core documentation (files in F<pod/>)
196
197=item 2
198
617eabfa
NC
199Work out how to split C<perlfunc> into chunks, preferably one per function
200group, preferably with general case code that could be used elsewhere.
201Challenges here are correctly identifying the groups of functions that go
202together, and making the right named external cross-links point to the right
203page. Things to be aware of are C<-X>, groups such as C<getpwnam> to
204C<endservent>, two or more C<=items> giving the different parameter lists, such
205as
adebf063
NC
206
207 =item substr EXPR,OFFSET,LENGTH,REPLACEMENT
adebf063 208 =item substr EXPR,OFFSET,LENGTH
adebf063
NC
209 =item substr EXPR,OFFSET
210
211and different parameter lists having different meanings. (eg C<select>)
212
213=back
3a89a73c 214
0bdfc961
NC
215=head2 compressed man pages
216
217Be able to install them. This would probably need a configure test to see how
218the system does compressed man pages (same directory/different directory?
219same filename/different filename), as well as tweaking the F<installman> script
220to compress as necessary.
221
30222c0f
NC
222=head2 Add a code coverage target to the Makefile
223
224Make it easy for anyone to run Devel::Cover on the core's tests. The steps
225to do this manually are roughly
226
227=over 4
228
229=item *
230
231do a normal C<Configure>, but include Devel::Cover as a module to install
232(see F<INSTALL> for how to do this)
233
234=item *
235
236 make perl
237
238=item *
239
240 cd t; HARNESS_PERL_SWITCHES=-MDevel::Cover ./perl -I../lib harness
241
242=item *
243
244Process the resulting Devel::Cover database
245
246=back
247
248This just give you the coverage of the F<.pm>s. To also get the C level
249coverage you need to
250
251=over 4
252
253=item *
254
255Additionally tell C<Configure> to use the appropriate C compiler flags for
256C<gcov>
257
258=item *
259
260 make perl.gcov
261
262(instead of C<make perl>)
263
264=item *
265
266After running the tests run C<gcov> to generate all the F<.gcov> files.
267(Including down in the subdirectories of F<ext/>
268
269=item *
270
271(From the top level perl directory) run C<gcov2perl> on all the C<.gcov> files
272to get their stats into the cover_db directory.
273
274=item *
275
276Then process the Devel::Cover database
277
278=back
279
280It would be good to add a single switch to C<Configure> to specify that you
281wanted to perform perl level coverage, and another to specify C level
282coverage, and have C<Configure> and the F<Makefile> do all the right things
283automatically.
284
02f21748 285=head2 Make Config.pm cope with differences between built and installed perl
0bdfc961
NC
286
287Quite often vendors ship a perl binary compiled with their (pay-for)
288compilers. People install a free compiler, such as gcc. To work out how to
289build extensions, Perl interrogates C<%Config>, so in this situation
290C<%Config> describes compilers that aren't there, and extension building
291fails. This forces people into choosing between re-compiling perl themselves
292using the compiler they have, or only using modules that the vendor ships.
293
294It would be good to find a way teach C<Config.pm> about the installation setup,
295possibly involving probing at install time or later, so that the C<%Config> in
296a binary distribution better describes the installed machine, when the
297installed machine differs from the build machine in some significant way.
298
728f4ecd
NC
299=head2 linker specification files
300
301Some platforms mandate that you provide a list of a shared library's external
302symbols to the linker, so the core already has the infrastructure in place to
303do this for generating shared perl libraries. My understanding is that the
304GNU toolchain can accept an optional linker specification file, and restrict
305visibility just to symbols declared in that file. It would be good to extend
306F<makedef.pl> to support this format, and to provide a means within
307C<Configure> to enable it. This would allow Unix users to test that the
308export list is correct, and to build a perl that does not pollute the global
309namespace with private symbols.
310
a229ae3b
RGS
311=head2 Cross-compile support
312
313Currently C<Configure> understands C<-Dusecrosscompile> option. This option
314arranges for building C<miniperl> for TARGET machine, so this C<miniperl> is
315assumed then to be copied to TARGET machine and used as a replacement of full
316C<perl> executable.
317
d1307786 318This could be done little differently. Namely C<miniperl> should be built for
a229ae3b 319HOST and then full C<perl> with extensions should be compiled for TARGET.
d1307786 320This, however, might require extra trickery for %Config: we have one config
87a942b1
JH
321first for HOST and then another for TARGET. Tools like MakeMaker will be
322mightily confused. Having around two different types of executables and
323libraries (HOST and TARGET) makes life interesting for Makefiles and
324shell (and Perl) scripts. There is $Config{run}, normally empty, which
325can be used as an execution wrapper. Also note that in some
326cross-compilation/execution environments the HOST and the TARGET do
327not see the same filesystem(s), the $Config{run} may need to do some
328file/directory copying back and forth.
0bdfc961
NC
329
330=head1 Tasks that need a little C knowledge
331
332These tasks would need a little C knowledge, but don't need any specific
333background or experience with XS, or how the Perl interpreter works
334
fbf638cb
RGS
335=head2 Modernize the order of directories in @INC
336
337The way @INC is laid out by default, one cannot upgrade core (dual-life)
338modules without overwriting files. This causes problems for binary
3d14fd97
AD
339package builders. One possible proposal is laid out in this
340message:
341L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2002-04/msg02380.html>.
fbf638cb 342
bcbaa2d5
RGS
343=head2 -Duse32bit*
344
345Natively 64-bit systems need neither -Duse64bitint nor -Duse64bitall.
346On these systems, it might be the default compilation mode, and there
347is currently no guarantee that passing no use64bitall option to the
348Configure process will build a 32bit perl. Implementing -Duse32bit*
349options would be nice for perl 5.12.
350
0bdfc961 351=head2 Make it clear from -v if this is the exact official release
89007cb3 352
617eabfa
NC
353Currently perl from C<p4>/C<rsync> ships with a F<patchlevel.h> file that
354usually defines one local patch, of the form "MAINT12345" or "RC1". The output
355of perl -v doesn't report that a perl isn't an official release, and this
89007cb3 356information can get lost in bugs reports. Because of this, the minor version
fa11829f 357isn't bumped up until RC time, to minimise the possibility of versions of perl
89007cb3
NC
358escaping that believe themselves to be newer than they actually are.
359
360It would be useful to find an elegant way to have the "this is an interim
361maintenance release" or "this is a release candidate" in the terse -v output,
362and have it so that it's easy for the pumpking to remove this just as the
363release tarball is rolled up. This way the version pulled out of rsync would
364always say "I'm a development release" and it would be safe to bump the
365reported minor version as soon as a release ships, which would aid perl
366developers.
367
0bdfc961
NC
368This task is really about thinking of an elegant way to arrange the C source
369such that it's trivial for the Pumpking to flag "this is an official release"
370when making a tarball, yet leave the default source saying "I'm not the
371official release".
372
fee0a0f7 373=head2 Profile Perl - am I hot or not?
62403a3c 374
fee0a0f7
NC
375The Perl source code is stable enough that it makes sense to profile it,
376identify and optimise the hotspots. It would be good to measure the
377performance of the Perl interpreter using free tools such as cachegrind,
378gprof, and dtrace, and work to reduce the bottlenecks they reveal.
379
380As part of this, the idea of F<pp_hot.c> is that it contains the I<hot> ops,
381the ops that are most commonly used. The idea is that by grouping them, their
382object code will be adjacent in the executable, so they have a greater chance
383of already being in the CPU cache (or swapped in) due to being near another op
384already in use.
62403a3c
NC
385
386Except that it's not clear if these really are the most commonly used ops. So
fee0a0f7
NC
387as part of exercising your skills with coverage and profiling tools you might
388want to determine what ops I<really> are the most commonly used. And in turn
389suggest evictions and promotions to achieve a better F<pp_hot.c>.
62403a3c 390
98fed0ad
NC
391=head2 Allocate OPs from arenas
392
393Currently all new OP structures are individually malloc()ed and free()d.
394All C<malloc> implementations have space overheads, and are now as fast as
395custom allocates so it would both use less memory and less CPU to allocate
396the various OP structures from arenas. The SV arena code can probably be
397re-used for this.
398
539f2c54
JC
399Note that Configuring perl with C<-Accflags=-DPL_OP_SLAB_ALLOC> will use
400Perl_Slab_alloc() to pack optrees into a contiguous block, which is
401probably superior to the use of OP arenas, esp. from a cache locality
402standpoint. See L<Profile Perl - am I hot or not?>.
403
a229ae3b 404=head2 Improve win32/wince.c
0bdfc961 405
a229ae3b 406Currently, numerous functions look virtually, if not completely,
02f21748 407identical in both C<win32/wince.c> and C<win32/win32.c> files, which can't
6d71adcd
NC
408be good.
409
c5b31784
SH
410=head2 Use secure CRT functions when building with VC8 on Win32
411
412Visual C++ 2005 (VC++ 8.x) deprecated a number of CRT functions on the basis
413that they were "unsafe" and introduced differently named secure versions of
414them as replacements, e.g. instead of writing
415
416 FILE* f = fopen(__FILE__, "r");
417
418one should now write
419
420 FILE* f;
421 errno_t err = fopen_s(&f, __FILE__, "r");
422
423Currently, the warnings about these deprecations have been disabled by adding
424-D_CRT_SECURE_NO_DEPRECATE to the CFLAGS. It would be nice to remove that
425warning suppressant and actually make use of the new secure CRT functions.
426
427There is also a similar issue with POSIX CRT function names like fileno having
428been deprecated in favour of ISO C++ conformant names like _fileno. These
26a6faa8 429warnings are also currently suppressed by adding -D_CRT_NONSTDC_NO_DEPRECATE. It
c5b31784
SH
430might be nice to do as Microsoft suggest here too, although, unlike the secure
431functions issue, there is presumably little or no benefit in this case.
432
6b632b43
NC
433=head2 __FUNCTION__ for MSVC-pre-7.0
434
435Jarkko notes that one can things morally equivalent to C<__FUNCTION__>
436(or C<__func__>) even in MSVC-pre-7.0, contrary to popular belief.
437See L<http://www.codeproject.com/debug/extendedtrace.asp> if you feel like
438making C<PERL_MEM_LOG> more useful on Win32.
439
6d71adcd
NC
440=head1 Tasks that need a knowledge of XS
441
442These tasks would need C knowledge, and roughly the level of knowledge of
443the perl API that comes from writing modules that use XS to interface to
444C.
445
6d71adcd
NC
446=head2 autovivification
447
448Make all autovivification consistent w.r.t LVALUE/RVALUE and strict/no strict;
449
450This task is incremental - even a little bit of work on it will help.
451
452=head2 Unicode in Filenames
453
454chdir, chmod, chown, chroot, exec, glob, link, lstat, mkdir, open,
455opendir, qx, readdir, readlink, rename, rmdir, stat, symlink, sysopen,
456system, truncate, unlink, utime, -X. All these could potentially accept
457Unicode filenames either as input or output (and in the case of system
458and qx Unicode in general, as input or output to/from the shell).
459Whether a filesystem - an operating system pair understands Unicode in
460filenames varies.
461
462Known combinations that have some level of understanding include
463Microsoft NTFS, Apple HFS+ (In Mac OS 9 and X) and Apple UFS (in Mac
464OS X), NFS v4 is rumored to be Unicode, and of course Plan 9. How to
465create Unicode filenames, what forms of Unicode are accepted and used
466(UCS-2, UTF-16, UTF-8), what (if any) is the normalization form used,
467and so on, varies. Finding the right level of interfacing to Perl
468requires some thought. Remember that an OS does not implicate a
469filesystem.
470
471(The Windows -C command flag "wide API support" has been at least
472temporarily retired in 5.8.1, and the -C has been repurposed, see
473L<perlrun>.)
474
87a942b1
JH
475Most probably the right way to do this would be this:
476L</"Virtualize operating system access">.
477
6d71adcd
NC
478=head2 Unicode in %ENV
479
480Currently the %ENV entries are always byte strings.
87a942b1 481See L</"Virtualize operating system access">.
6d71adcd 482
1f2e7916
JD
483=head2 Unicode and glob()
484
485Currently glob patterns and filenames returned from File::Glob::glob()
87a942b1 486are always byte strings. See L</"Virtualize operating system access">.
1f2e7916 487
dbb0c492
RGS
488=head2 Unicode and lc/uc operators
489
490Some built-in operators (C<lc>, C<uc>, etc.) behave differently, based on
491what the internal encoding of their argument is. That should not be the
492case. Maybe add a pragma to switch behaviour.
493
6d71adcd
NC
494=head2 use less 'memory'
495
496Investigate trade offs to switch out perl's choices on memory usage.
497Particularly perl should be able to give memory back.
498
499This task is incremental - even a little bit of work on it will help.
500
501=head2 Re-implement C<:unique> in a way that is actually thread-safe
502
503The old implementation made bad assumptions on several levels. A good 90%
504solution might be just to make C<:unique> work to share the string buffer
505of SvPVs. That way large constant strings can be shared between ithreads,
506such as the configuration information in F<Config>.
507
508=head2 Make tainting consistent
509
510Tainting would be easier to use if it didn't take documented shortcuts and
511allow taint to "leak" everywhere within an expression.
512
513=head2 readpipe(LIST)
514
515system() accepts a LIST syntax (and a PROGRAM LIST syntax) to avoid
516running a shell. readpipe() (the function behind qx//) could be similarly
517extended.
518
519=head2 strcat(), strcpy(), strncat(), strncpy(), sprintf(), vsprintf()
520
521Maybe create a utility that checks after each libperl.a creation that
522none of the above (nor sprintf(), vsprintf(), or *SHUDDER* gets())
523ever creep back to libperl.a.
524
525 nm libperl.a | ./miniperl -alne '$o = $F[0] if /:$/; print "$o $F[1]" if $F[0] eq "U" && $F[1] =~ /^(?:strn?c(?:at|py)|v?sprintf|gets)$/'
526
527Note, of course, that this will only tell whether B<your> platform
528is using those naughty interfaces.
529
530=head2 Audit the code for destruction ordering assumptions
531
532Change 25773 notes
533
534 /* Need to check SvMAGICAL, as during global destruction it may be that
535 AvARYLEN(av) has been freed before av, and hence the SvANY() pointer
536 is now part of the linked list of SV heads, rather than pointing to
537 the original body. */
538 /* FIXME - audit the code for other bugs like this one. */
539
540adding the C<SvMAGICAL> check to
541
542 if (AvARYLEN(av) && SvMAGICAL(AvARYLEN(av))) {
543 MAGIC *mg = mg_find (AvARYLEN(av), PERL_MAGIC_arylen);
544
545Go through the core and look for similar assumptions that SVs have particular
546types, as all bets are off during global destruction.
547
749904bf
JH
548=head2 Extend PerlIO and PerlIO::Scalar
549
550PerlIO::Scalar doesn't know how to truncate(). Implementing this
551would require extending the PerlIO vtable.
552
553Similarly the PerlIO vtable doesn't know about formats (write()), or
554about stat(), or chmod()/chown(), utime(), or flock().
555
556(For PerlIO::Scalar it's hard to see what e.g. mode bits or ownership
557would mean.)
558
559PerlIO doesn't do directories or symlinks, either: mkdir(), rmdir(),
560opendir(), closedir(), seekdir(), rewinddir(), glob(); symlink(),
561readlink().
562
94da6c29
JH
563See also L</"Virtualize operating system access">.
564
3236f110
NC
565=head2 -C on the #! line
566
567It should be possible to make -C work correctly if found on the #! line,
568given that all perl command line options are strict ASCII, and -C changes
569only the interpretation of non-ASCII characters, and not for the script file
570handle. To make it work needs some investigation of the ordering of function
571calls during startup, and (by implication) a bit of tweaking of that order.
572
573
0bdfc961 574=head1 Tasks that need a knowledge of the interpreter
3298bd4d 575
0bdfc961
NC
576These tasks would need C knowledge, and knowledge of how the interpreter works,
577or a willingness to learn.
3298bd4d 578
636e63cb
NC
579=head2 state variable initialization in list context
580
581Currently this is illegal:
582
583 state ($a, $b) = foo();
584
585The current Perl 6 design is that C<state ($a) = foo();> and
586C<(state $a) = foo();> have different semantics, which is tricky to implement
587in Perl 5 as currently the produce the same opcode trees. It would be useful
588to clarify that the Perl 6 design is firm, and then implement the necessary
589code in Perl 5. There are comments in C<Perl_newASSIGNOP()> that show the
590code paths taken by various assignment constructions involving state variables.
591
4fedb12c
RGS
592=head2 Implement $value ~~ 0 .. $range
593
594It would be nice to extend the syntax of the C<~~> operator to also
595understand numeric (and maybe alphanumeric) ranges.
a393eb28
RGS
596
597=head2 A does() built-in
598
599Like ref(), only useful. It would call the C<DOES> method on objects; it
600would also tell whether something can be dereferenced as an
601array/hash/etc., or used as a regexp, etc.
602L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-03/msg00481.html>
603
604=head2 Tied filehandles and write() don't mix
605
606There is no method on tied filehandles to allow them to be called back by
607formats.
4fedb12c 608
d10fc472 609=head2 Attach/detach debugger from running program
1626a787 610
cd793d32
NC
611The old perltodo notes "With C<gdb>, you can attach the debugger to a running
612program if you pass the process ID. It would be good to do this with the Perl
0bdfc961
NC
613debugger on a running Perl program, although I'm not sure how it would be
614done." ssh and screen do this with named pipes in /tmp. Maybe we can too.
1626a787 615
a8cb5b9e
RGS
616=head2 Optimize away empty destructors
617
618Defining an empty DESTROY method might be useful (notably in
619AUTOLOAD-enabled classes), but it's still a bit expensive to call. That
620could probably be optimized.
621
0bdfc961
NC
622=head2 LVALUE functions for lists
623
624The old perltodo notes that lvalue functions don't work for list or hash
625slices. This would be good to fix.
626
627=head2 LVALUE functions in the debugger
628
629The old perltodo notes that lvalue functions don't work in the debugger. This
630would be good to fix.
631
0bdfc961
NC
632=head2 regexp optimiser optional
633
634The regexp optimiser is not optional. It should configurable to be, to allow
635its performance to be measured, and its bugs to be easily demonstrated.
636
02f21748
RGS
637=head2 delete &function
638
639Allow to delete functions. One can already undef them, but they're still
640in the stash.
641
ef36c6a7
RGS
642=head2 C</w> regex modifier
643
644That flag would enable to match whole words, and also to interpolate
645arrays as alternations. With it, C</P/w> would be roughly equivalent to:
646
647 do { local $"='|'; /\b(?:P)\b/ }
648
649See L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-01/msg00400.html>
650for the discussion.
651
0bdfc961
NC
652=head2 optional optimizer
653
654Make the peephole optimizer optional. Currently it performs two tasks as
655it walks the optree - genuine peephole optimisations, and necessary fixups of
656ops. It would be good to find an efficient way to switch out the
657optimisations whilst keeping the fixups.
658
659=head2 You WANT *how* many
660
661Currently contexts are void, scalar and list. split has a special mechanism in
662place to pass in the number of return values wanted. It would be useful to
663have a general mechanism for this, backwards compatible and little speed hit.
664This would allow proposals such as short circuiting sort to be implemented
665as a module on CPAN.
666
667=head2 lexical aliases
668
669Allow lexical aliases (maybe via the syntax C<my \$alias = \$foo>.
670
671=head2 entersub XS vs Perl
672
673At the moment pp_entersub is huge, and has code to deal with entering both
674perl and XS subroutines. Subroutine implementations rarely change between
675perl and XS at run time, so investigate using 2 ops to enter subs (one for
676XS, one for perl) and swap between if a sub is redefined.
2810d901
NC
677
678=head2 Self ties
679
680self ties are currently illegal because they caused too many segfaults. Maybe
681the causes of these could be tracked down and self-ties on all types re-
682instated.
0bdfc961
NC
683
684=head2 Optimize away @_
685
686The old perltodo notes "Look at the "reification" code in C<av.c>".
687
16fc99ce
NC
688=head2 Properly Unicode safe tokeniser and pads.
689
690The tokeniser isn't actually very UTF-8 clean. C<use utf8;> is a hack -
691variable names are stored in stashes as raw bytes, without the utf-8 flag
692set. The pad API only takes a C<char *> pointer, so that's all bytes too. The
693tokeniser ignores the UTF-8-ness of C<PL_rsfp>, or any SVs returned from
694source filters. All this could be fixed.
695
f092b1f4
RGS
696=head2 The yada yada yada operators
697
698Perl 6's Synopsis 3 says:
699
700I<The ... operator is the "yada, yada, yada" list operator, which is used as
701the body in function prototypes. It complains bitterly (by calling fail)
702if it is ever executed. Variant ??? calls warn, and !!! calls die.>
703
704Those would be nice to add to Perl 5. That could be done without new ops.
705
87a942b1
JH
706=head2 Virtualize operating system access
707
708Implement a set of "vtables" that virtualizes operating system access
709(open(), mkdir(), unlink(), readdir(), getenv(), etc.) At the very
710least these interfaces should take SVs as "name" arguments instead of
711bare char pointers; probably the most flexible and extensible way
e1a3d5d1
JH
712would be for the Perl-facing interfaces to accept HVs. The system
713needs to be per-operating-system and per-file-system
714hookable/filterable, preferably both from XS and Perl level
87a942b1
JH
715(L<perlport/"Files and Filesystems"> is good reading at this point,
716in fact, all of L<perlport> is.)
717
e1a3d5d1
JH
718This has actually already been implemented (but only for Win32),
719take a look at F<iperlsys.h> and F<win32/perlhost.h>. While all Win32
720variants go through a set of "vtables" for operating system access,
721non-Win32 systems currently go straight for the POSIX/UNIX-style
722system/library call. Similar system as for Win32 should be
723implemented for all platforms. The existing Win32 implementation
724probably does not need to survive alongside this proposed new
725implementation, the approaches could be merged.
87a942b1
JH
726
727What would this give us? One often-asked-for feature this would
94da6c29
JH
728enable is using Unicode for filenames, and other "names" like %ENV,
729usernames, hostnames, and so forth.
730(See L<perlunicode/"When Unicode Does Not Happen">.)
731
732But this kind of virtualization would also allow for things like
733virtual filesystems, virtual networks, and "sandboxes" (though as long
734as dynamic loading of random object code is allowed, not very safe
735sandboxes since external code of course know not of Perl's vtables).
736An example of a smaller "sandbox" is that this feature can be used to
737implement per-thread working directories: Win32 already does this.
738
739See also L</"Extend PerlIO and PerlIO::Scalar">.
87a942b1 740
0bdfc961
NC
741=head1 Big projects
742
743Tasks that will get your name mentioned in the description of the "Highlights
87a942b1 744of 5.12"
0bdfc961
NC
745
746=head2 make ithreads more robust
747
4e577f8b 748Generally make ithreads more robust. See also L</iCOW>
0bdfc961
NC
749
750This task is incremental - even a little bit of work on it will help, and
751will be greatly appreciated.
752
6c047da7
YST
753One bit would be to write the missing code in sv.c:Perl_dirp_dup.
754
59c7f7d5
RGS
755Fix Perl_sv_dup, et al so that threads can return objects.
756
0bdfc961
NC
757=head2 iCOW
758
759Sarathy and Arthur have a proposal for an improved Copy On Write which
760specifically will be able to COW new ithreads. If this can be implemented
761it would be a good thing.
762
763=head2 (?{...}) closures in regexps
764
765Fix (or rewrite) the implementation of the C</(?{...})/> closures.
766
767=head2 A re-entrant regexp engine
768
769This will allow the use of a regex from inside (?{ }), (??{ }) and
770(?(?{ })|) constructs.
6bda09f9 771
6bda09f9
YO
772=head2 Add class set operations to regexp engine
773
774Apparently these are quite useful. Anyway, Jeffery Friedl wants them.
775
776demerphq has this on his todo list, but right at the bottom.