7 This is a list of wishes for Perl. The most up to date version of this file
8 is at L<http://perl5.git.perl.org/perl.git/blob_plain/HEAD:/Porting/todo.pod>
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
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
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/>
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?
26 =head1 Tasks that only need Perl knowledge
28 =head2 Migrate t/ from custom TAP generation
30 Many tests below F<t/> still generate TAP by "hand", rather than using library
31 functions. As explained in L<perlhack/TESTING>, tests in F<t/> are
32 written in a particular way to test that more complex constructions actually
33 work before using them routinely. Hence they don't use C<Test::More>, but
34 instead there is an intentionally simpler library, F<t/test.pl>. However,
35 quite a few tests in F<t/> have not been refactored to use it. Refactoring
36 any of these tests, one at a time, is a useful thing TODO.
38 The subdirectories F<base>, F<cmd> and F<comp>, that contain the most
39 basic tests, should be excluded from this task.
41 =head2 Automate perldelta generation
43 The perldelta file accompanying each release summaries the major changes.
44 It's mostly manually generated currently, but some of that could be
45 automated with a bit of perl, specifically the generation of
49 =item Modules and Pragmata
51 =item New Documentation
57 See F<Porting/how_to_write_a_perldelta.pod> for details.
59 =head2 Make Schwern poorer
61 We should have tests for everything. When all the core's modules are tested,
62 Schwern has promised to donate to $500 to TPF. We may need volunteers to
63 hold him upside down and shake vigorously in order to actually extract the
66 =head2 Improve the coverage of the core tests
68 Use Devel::Cover to ascertain the core modules' test coverage, then add
69 tests that are currently missing.
73 A full test suite for the B module would be nice.
75 =head2 A decent benchmark
77 C<perlbench> seems impervious to any recent changes made to the perl core. It
78 would be useful to have a reasonable general benchmarking suite that roughly
79 represented what current perl programs do, and measurably reported whether
80 tweaks to the core improve, degrade or don't really affect performance, to
81 guide people attempting to optimise the guts of perl. Gisle would welcome
82 new tests for perlbench. Steffen Schwingon would welcome help with
83 L<Benchmark::Perl::Formance>
85 =head2 fix tainting bugs
87 Fix the bugs revealed by running the test suite with the C<-t> switch (via
88 C<make test.taintwarn>).
90 =head2 Dual life everything
92 As part of the "dists" plan, anything that doesn't belong in the smallest perl
93 distribution needs to be dual lifed. Anything else can be too. Figure out what
94 changes would be needed to package that module and its tests up for CPAN, and
95 do so. Test it with older perl releases, and fix the problems you find.
97 To make a minimal perl distribution, it's useful to look at
98 F<t/lib/commonsense.t>.
100 =head2 POSIX memory footprint
102 Ilya observed that use POSIX; eats memory like there's no tomorrow, and at
103 various times worked to cut it down. There is probably still fat to cut out -
104 for example POSIX passes Exporter some very memory hungry data structures.
106 =head2 makedef.pl and conditional compilation
108 The script F<makedef.pl> that generates the list of exported symbols on
109 platforms which need this. Functions are declared in F<embed.fnc>, variables
110 in F<intrpvar.h>. Quite a few of the functions and variables are conditionally
111 declared there, using C<#ifdef>. However, F<makedef.pl> doesn't understand the
112 C macros, so the rules about which symbols are present when is duplicated in
113 the Perl code. Writing things twice is bad, m'kay. It would be good to teach
114 F<.pl> to understand the conditional compilation, and hence remove the
115 duplication, and the mistakes it has caused.
117 =head2 use strict; and AutoLoad
119 Currently if you write
122 use AutoLoader 'AUTOLOAD';
127 print join (' ', No, strict, here), "!\n";
130 then C<use strict;> isn't in force within the autoloaded subroutines. It would
131 be more consistent (and less surprising) to arrange for all lexical pragmas
132 in force at the __END__ block to be in force within each autoloaded subroutine.
134 There's a similar problem with SelfLoader.
136 =head2 profile installman
138 The F<installman> script is slow. All it is doing text processing, which we're
139 told is something Perl is good at. So it would be nice to know what it is doing
140 that is taking so much CPU, and where possible address it.
142 =head2 enable lexical enabling/disabling of individual warnings
144 Currently, warnings can only be enabled or disabled by category. There
145 are times when it would be useful to quash a single warning, not a
148 =head2 document diagnostics
150 Many diagnostic messages are not currently documented. The list is at the end
153 =head1 Tasks that need a little sysadmin-type knowledge
155 Or if you prefer, tasks that you would learn from, and broaden your skills
158 =head2 make HTML install work
160 There is an C<install.html> target in the Makefile. It's marked as
161 "experimental". It would be good to get this tested, make it work reliably, and
162 remove the "experimental" tag. This would include
168 Checking that cross linking between various parts of the documentation works.
169 In particular that links work between the modules (files with POD in F<lib/>)
170 and the core documentation (files in F<pod/>)
174 Improving the code that split C<perlfunc> into chunks, preferably with
175 general case code added to L<Pod::Functions> that could be used elsewhere.
177 Challenges here are correctly identifying the groups of functions that go
178 together, and making the right named external cross-links point to the right
179 page. Currently this works reasonably well in the general case, and correctly
180 parses two or more C<=items> giving the different parameter lists for the
181 same function, such used by C<substr>. However it fails completely where
182 I<different> functions are listed as a sequence of C<=items> but share the
183 same description. All the functions from C<getpwnam> to C<endprotoent> have
184 individual stub pages, with only the page for C<endservent> holding the
185 description common to all. Likewise C<q>, C<qq> and C<qw> have stub pages,
186 instead of sharing the body of C<qx>.
188 Note also the current code isn't ideal with the two forms of C<select>, mushing
189 them both into one F<select.html> with the two descriptions run together.
190 Fixing this may well be a special case.
194 =head2 compressed man pages
196 Be able to install them. This would probably need a configure test to see how
197 the system does compressed man pages (same directory/different directory?
198 same filename/different filename), as well as tweaking the F<installman> script
199 to compress as necessary.
201 =head2 Add a code coverage target to the Makefile
203 Make it easy for anyone to run Devel::Cover on the core's tests. The steps
204 to do this manually are roughly
210 do a normal C<Configure>, but include Devel::Cover as a module to install
211 (see L<INSTALL> for how to do this)
219 cd t; HARNESS_PERL_SWITCHES=-MDevel::Cover ./perl -I../lib harness
223 Process the resulting Devel::Cover database
227 This just give you the coverage of the F<.pm>s. To also get the C level
234 Additionally tell C<Configure> to use the appropriate C compiler flags for
241 (instead of C<make perl>)
245 After running the tests run C<gcov> to generate all the F<.gcov> files.
246 (Including down in the subdirectories of F<ext/>
250 (From the top level perl directory) run C<gcov2perl> on all the C<.gcov> files
251 to get their stats into the cover_db directory.
255 Then process the Devel::Cover database
259 It would be good to add a single switch to C<Configure> to specify that you
260 wanted to perform perl level coverage, and another to specify C level
261 coverage, and have C<Configure> and the F<Makefile> do all the right things
264 =head2 Make Config.pm cope with differences between built and installed perl
266 Quite often vendors ship a perl binary compiled with their (pay-for)
267 compilers. People install a free compiler, such as gcc. To work out how to
268 build extensions, Perl interrogates C<%Config>, so in this situation
269 C<%Config> describes compilers that aren't there, and extension building
270 fails. This forces people into choosing between re-compiling perl themselves
271 using the compiler they have, or only using modules that the vendor ships.
273 It would be good to find a way teach C<Config.pm> about the installation setup,
274 possibly involving probing at install time or later, so that the C<%Config> in
275 a binary distribution better describes the installed machine, when the
276 installed machine differs from the build machine in some significant way.
278 =head2 linker specification files
280 Some platforms mandate that you provide a list of a shared library's external
281 symbols to the linker, so the core already has the infrastructure in place to
282 do this for generating shared perl libraries. Florian Ragwitz has been working
283 to offer this for the GNU toolchain, to allow Unix users to test that the
284 export list is correct, and to build a perl that does not pollute the global
285 namespace with private symbols, and will fail in the same way as msvc or mingw
286 builds or when using PERL_DL_NONLAZY=1. See the branch smoke-me/rafl/ld_export
288 =head2 Cross-compile support
290 We get requests for "how to cross compile Perl". The vast majority of these
291 seem to be for a couple of scenarios:
297 Platforms that could build natively using F<./Configure> (I<e.g.> Linux or
298 NetBSD on MIPS or ARM) but people want to use a beefier machine (and on the
299 same OS) to build more easily.
303 Platforms that can't build natively, but no (significant) porting changes
304 are needed to our current source code. Prime example of this is Android.
308 There are several scripts and tools for cross-compiling perl for other
309 platforms. However, these are somewhat inconsistent and scattered across the
310 codebase, none are documented well, none are clearly flexible enough to
311 be confident that they can support any TARGET/HOST plaform pair other than
312 that which they were developed on, and it's not clear how bitrotted they are.
314 For example, C<Configure> understands C<-Dusecrosscompile> option. This option
315 arranges for building C<miniperl> for TARGET machine, so this C<miniperl> is
316 assumed then to be copied to TARGET machine and used as a replacement of
317 full C<perl> executable. This code is almost 10 years old. Meanwhile, the
318 F<Cross/> directory contains two different approaches for cross compiling to
319 ARM Linux targets, relying on hand curated F<config.sh> files, but that code
320 is getting on for 5 years old, and requires insider knowledge of perl's
321 build system to draft a F<config.sh> for a new platform.
323 Jess Robinson has sumbitted a grant to TPF to work on cleaning this up.
325 =head2 Split "linker" from "compiler"
327 Right now, Configure probes for two commands, and sets two variables:
331 =item * C<cc> (in F<cc.U>)
333 This variable holds the name of a command to execute a C compiler which
334 can resolve multiple global references that happen to have the same
335 name. Usual values are F<cc> and F<gcc>.
336 Fervent ANSI compilers may be called F<c89>. AIX has F<xlc>.
338 =item * C<ld> (in F<dlsrc.U>)
340 This variable indicates the program to be used to link
341 libraries for dynamic loading. On some systems, it is F<ld>.
342 On ELF systems, it should be C<$cc>. Mostly, we'll try to respect
343 the hint file setting.
347 There is an implicit historical assumption from around Perl5.000alpha
348 something, that C<$cc> is also the correct command for linking object files
349 together to make an executable. This may be true on Unix, but it's not true
350 on other platforms, and there are a maze of work arounds in other places (such
351 as F<Makefile.SH>) to cope with this.
353 Ideally, we should create a new variable to hold the name of the executable
354 linker program, probe for it in F<Configure>, and centralise all the special
355 case logic there or in hints files.
357 A small bikeshed issue remains - what to call it, given that C<$ld> is already
358 taken (arguably for the wrong thing now, but on SunOS 4.1 it is the command
359 for creating dynamically-loadable modules) and C<$link> could be confused with
360 the Unix command line executable of the same name, which does something
361 completely different. Andy Dougherty makes the counter argument "In parrot, I
362 tried to call the command used to link object files and libraries into an
363 executable F<link>, since that's what my vaguely-remembered DOS and VMS
364 experience suggested. I don't think any real confusion has ensued, so it's
365 probably a reasonable name for perl5 to use."
367 "Alas, I've always worried that introducing it would make things worse,
368 since now the module building utilities would have to look for
369 C<$Config{link}> and institute a fall-back plan if it weren't found."
370 Although I can see that as confusing, given that C<$Config{d_link}> is true
371 when (hard) links are available.
373 =head2 Configure Windows using PowerShell
375 Currently, Windows uses hard-coded config files based to build the
376 config.h for compiling Perl. Makefiles are also hard-coded and need to be
377 hand edited prior to building Perl. While this makes it easy to create a perl.exe
378 that works across multiple Windows versions, being able to accurately
379 configure a perl.exe for a specific Windows versions and VS C++ would be
380 a nice enhancement. With PowerShell available on Windows XP and up, this
381 may now be possible. Step 1 might be to investigate whether this is possible
382 and use this to clean up our current makefile situation. Step 2 would be to
383 see if there would be a way to use our existing metaconfig units to configure a
384 Windows Perl or whether we go in a separate direction and make it so. Of
385 course, we all know what step 3 is.
387 =head1 Tasks that need a little C knowledge
389 These tasks would need a little C knowledge, but don't need any specific
390 background or experience with XS, or how the Perl interpreter works
392 =head2 Weed out needless PERL_UNUSED_ARG
394 The C code uses the macro C<PERL_UNUSED_ARG> to stop compilers warning about
395 unused arguments. Often the arguments can't be removed, as there is an
396 external constraint that determines the prototype of the function, so this
397 approach is valid. However, there are some cases where C<PERL_UNUSED_ARG>
398 could be removed. Specifically
404 The prototypes of (nearly all) static functions can be changed
408 Unused arguments generated by short cut macros are wasteful - the short cut
409 macro used can be changed.
415 Natively 64-bit systems need neither -Duse64bitint nor -Duse64bitall.
416 On these systems, it might be the default compilation mode, and there
417 is currently no guarantee that passing no use64bitall option to the
418 Configure process will build a 32bit perl. Implementing -Duse32bit*
419 options would be nice for perl 5.18.0.
421 =head2 Profile Perl - am I hot or not?
423 The Perl source code is stable enough that it makes sense to profile it,
424 identify and optimise the hotspots. It would be good to measure the
425 performance of the Perl interpreter using free tools such as cachegrind,
426 gprof, and dtrace, and work to reduce the bottlenecks they reveal.
428 As part of this, the idea of F<pp_hot.c> is that it contains the I<hot> ops,
429 the ops that are most commonly used. The idea is that by grouping them, their
430 object code will be adjacent in the executable, so they have a greater chance
431 of already being in the CPU cache (or swapped in) due to being near another op
434 Except that it's not clear if these really are the most commonly used ops. So
435 as part of exercising your skills with coverage and profiling tools you might
436 want to determine what ops I<really> are the most commonly used. And in turn
437 suggest evictions and promotions to achieve a better F<pp_hot.c>.
439 One piece of Perl code that might make a good testbed is F<installman>.
441 =head2 Allocate OPs from arenas
443 Currently all new OP structures are individually malloc()ed and free()d.
444 All C<malloc> implementations have space overheads, and are now as fast as
445 custom allocates so it would both use less memory and less CPU to allocate
446 the various OP structures from arenas. The SV arena code can probably be
449 Note that Configuring perl with C<-Accflags=-DPL_OP_SLAB_ALLOC> will use
450 Perl_Slab_alloc() to pack optrees into a contiguous block, which is
451 probably superior to the use of OP arenas, esp. from a cache locality
452 standpoint. See L<Profile Perl - am I hot or not?>.
454 =head2 Improve win32/wince.c
456 Currently, numerous functions look virtually, if not completely,
457 identical in both F<win32/wince.c> and F<win32/win32.c> files, which can't
460 =head2 Use secure CRT functions when building with VC8 on Win32
462 Visual C++ 2005 (VC++ 8.x) deprecated a number of CRT functions on the basis
463 that they were "unsafe" and introduced differently named secure versions of
464 them as replacements, e.g. instead of writing
466 FILE* f = fopen(__FILE__, "r");
471 errno_t err = fopen_s(&f, __FILE__, "r");
473 Currently, the warnings about these deprecations have been disabled by adding
474 -D_CRT_SECURE_NO_DEPRECATE to the CFLAGS. It would be nice to remove that
475 warning suppressant and actually make use of the new secure CRT functions.
477 There is also a similar issue with POSIX CRT function names like fileno having
478 been deprecated in favour of ISO C++ conformant names like _fileno. These
479 warnings are also currently suppressed by adding -D_CRT_NONSTDC_NO_DEPRECATE. It
480 might be nice to do as Microsoft suggest here too, although, unlike the secure
481 functions issue, there is presumably little or no benefit in this case.
483 =head2 Fix POSIX::access() and chdir() on Win32
485 These functions currently take no account of DACLs and therefore do not behave
486 correctly in situations where access is restricted by DACLs (as opposed to the
487 read-only attribute).
489 Furthermore, POSIX::access() behaves differently for directories having the
490 read-only attribute set depending on what CRT library is being used. For
491 example, the _access() function in the VC6 and VC7 CRTs (wrongly) claim that
492 such directories are not writable, whereas in fact all directories are writable
493 unless access is denied by DACLs. (In the case of directories, the read-only
494 attribute actually only means that the directory cannot be deleted.) This CRT
495 bug is fixed in the VC8 and VC9 CRTs (but, of course, the directory may still
496 not actually be writable if access is indeed denied by DACLs).
498 For the chdir() issue, see ActiveState bug #74552:
499 L<http://bugs.activestate.com/show_bug.cgi?id=74552>
501 Therefore, DACLs should be checked both for consistency across CRTs and for
504 (Note that perl's -w operator should not be modified to check DACLs. It has
505 been written so that it reflects the state of the read-only attribute, even
506 for directories (whatever CRT is being used), for symmetry with chmod().)
508 =head2 strcat(), strcpy(), strncat(), strncpy(), sprintf(), vsprintf()
510 Maybe create a utility that checks after each libperl.a creation that
511 none of the above (nor sprintf(), vsprintf(), or *SHUDDER* gets())
512 ever creep back to libperl.a.
514 nm libperl.a | ./miniperl -alne '$o = $F[0] if /:$/; print "$o $F[1]" if $F[0] eq "U" && $F[1] =~ /^(?:strn?c(?:at|py)|v?sprintf|gets)$/'
516 Note, of course, that this will only tell whether B<your> platform
517 is using those naughty interfaces.
519 =head2 -D_FORTIFY_SOURCE=2
521 Recent glibcs support C<-D_FORTIFY_SOURCE=2> which gives
522 protection against various kinds of buffer overflow problems.
523 It should probably be used for compiling Perl whenever available,
524 Configure and/or hints files should be adjusted to probe for the
525 availability of these feature and enable it as appropriate.
527 =head2 Arenas for GPs? For MAGIC?
529 C<struct gp> and C<struct magic> are both currently allocated by C<malloc>.
530 It might be a speed or memory saving to change to using arenas. Or it might
531 not. It would need some suitable benchmarking first. In particular, C<GP>s
532 can probably be changed with minimal compatibility impact (probably nothing
533 outside of the core, or even outside of F<gv.c> allocates them), but they
534 probably aren't allocated/deallocated often enough for a speed saving. Whereas
535 C<MAGIC> is allocated/deallocated more often, but in turn, is also something
536 more externally visible, so changing the rules here may bite external code.
540 Several SV body structs are now the same size, notably PVMG and PVGV, PVAV and
541 PVHV, and PVCV and PVFM. It should be possible to allocate and return same
542 sized bodies from the same actual arena, rather than maintaining one arena for
543 each. This could save 4-6K per thread, of memory no longer tied up in the
544 not-yet-allocated part of an arena.
547 =head1 Tasks that need a knowledge of XS
549 These tasks would need C knowledge, and roughly the level of knowledge of
550 the perl API that comes from writing modules that use XS to interface to
553 =head2 Write an XS cookbook
555 Create pod/perlxscookbook.pod with short, task-focused 'recipes' in XS that
556 demonstrate common tasks and good practices. (Some of these might be
557 extracted from perlguts.) The target audience should be XS novices, who need
558 more examples than perlguts but something less overwhelming than perlapi.
559 Recipes should provide "one pretty good way to do it" instead of TIMTOWTDI.
561 Rather than focusing on interfacing Perl to C libraries, such a cookbook
562 should probably focus on how to optimize Perl routines by re-writing them
563 in XS. This will likely be more motivating to those who mostly work in
564 Perl but are looking to take the next step into XS.
566 Deconstructing and explaining some simpler XS modules could be one way to
567 bootstrap a cookbook. (List::Util? Class::XSAccessor? Tree::Ternary_XS?)
568 Another option could be deconstructing the implementation of some simpler
571 =head2 Document how XSUBs can use C<cv_set_call_checker> to inline themselves as OPs
573 For a simple XSUB, often the subroutine dispatch takes more time than the
574 XSUB itself. v5.14.0 now allows XSUBs to register a function which will be
575 called when the parser is finished building an C<entersub> op which calls
578 Registration is done with C<Perl_cv_set_call_checker>, is documented at the
579 API level in L<perlapi>, and L<perl5140delta/Custom per-subroutine check hooks>
580 notes that it can be used to inline a subroutine, by replacing it with a
581 custom op. However there is no further detail of the code needed to do this.
582 It would be useful to add one or more annotated examples of how to create
585 This should provide a measurable speed up to simple XSUBs inside
586 tight loops. Initially one would have to write the OP alternative
587 implementation by hand, but it's likely that this should be reasonably
588 straightforward for the type of XSUB that would benefit the most. Longer
589 term, once the run-time implementation is proven, it should be possible to
590 progressively update ExtUtils::ParseXS to generate OP implementations for
593 =head2 Remove the use of SVs as temporaries in dump.c
595 F<dump.c> contains debugging routines to dump out the contains of perl data
596 structures, such as C<SV>s, C<AV>s and C<HV>s. Currently, the dumping code
597 B<uses> C<SV>s for its temporary buffers, which was a logical initial
598 implementation choice, as they provide ready made memory handling.
600 However, they also lead to a lot of confusion when it happens that what you're
601 trying to debug is seen by the code in F<dump.c>, correctly or incorrectly, as
602 a temporary scalar it can use for a temporary buffer. It's also not possible
603 to dump scalars before the interpreter is properly set up, such as during
604 ithreads cloning. It would be good to progressively replace the use of scalars
605 as string accumulation buffers with something much simpler, directly allocated
606 by C<malloc>. The F<dump.c> code is (or should be) only producing 7 bit
607 US-ASCII, so output character sets are not an issue.
609 Producing and proving an internal simple buffer allocation would make it easier
610 to re-write the internals of the PerlIO subsystem to avoid using C<SV>s for
611 B<its> buffers, use of which can cause problems similar to those of F<dump.c>,
614 =head2 safely supporting POSIX SA_SIGINFO
616 Some years ago Jarkko supplied patches to provide support for the POSIX
617 SA_SIGINFO feature in Perl, passing the extra data to the Perl signal handler.
619 Unfortunately, it only works with "unsafe" signals, because under safe
620 signals, by the time Perl gets to run the signal handler, the extra
621 information has been lost. Moreover, it's not easy to store it somewhere,
622 as you can't call mutexs, or do anything else fancy, from inside a signal
625 So it strikes me that we could provide safe SA_SIGINFO support
631 Provide global variables for two file descriptors
635 When the first request is made via C<sigaction> for C<SA_SIGINFO>, create a
636 pipe, store the reader in one, the writer in the other
640 In the "safe" signal handler (C<Perl_csighandler()>/C<S_raise_signal()>), if
641 the C<siginfo_t> pointer non-C<NULL>, and the writer file handle is open,
647 serialise signal number, C<struct siginfo_t> (or at least the parts we care
648 about) into a small auto char buff
652 C<write()> that (non-blocking) to the writer fd
658 if it writes 100%, flag the signal in a counter of "signals on the pipe" akin
659 to the current per-signal-number counts
663 if it writes 0%, assume the pipe is full. Flag the data as lost?
667 if it writes partially, croak a panic, as your OS is broken.
675 in the regular C<PERL_ASYNC_CHECK()> processing, if there are "signals on
676 the pipe", read the data out, deserialise, build the Perl structures on
677 the stack (code in C<Perl_sighandler()>, the "unsafe" handler), and call as
682 I think that this gets us decent C<SA_SIGINFO> support, without the current risk
683 of running Perl code inside the signal handler context. (With all the dangers
684 of things like C<malloc> corruption that that currently offers us)
686 For more information see the thread starting with this message:
687 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-03/msg00305.html>
689 =head2 autovivification
691 Make all autovivification consistent w.r.t LVALUE/RVALUE and strict/no strict;
693 This task is incremental - even a little bit of work on it will help.
695 =head2 Unicode in Filenames
697 chdir, chmod, chown, chroot, exec, glob, link, lstat, mkdir, open,
698 opendir, qx, readdir, readlink, rename, rmdir, stat, symlink, sysopen,
699 system, truncate, unlink, utime, -X. All these could potentially accept
700 Unicode filenames either as input or output (and in the case of system
701 and qx Unicode in general, as input or output to/from the shell).
702 Whether a filesystem - an operating system pair understands Unicode in
705 Known combinations that have some level of understanding include
706 Microsoft NTFS, Apple HFS+ (In Mac OS 9 and X) and Apple UFS (in Mac
707 OS X), NFS v4 is rumored to be Unicode, and of course Plan 9. How to
708 create Unicode filenames, what forms of Unicode are accepted and used
709 (UCS-2, UTF-16, UTF-8), what (if any) is the normalization form used,
710 and so on, varies. Finding the right level of interfacing to Perl
711 requires some thought. Remember that an OS does not implicate a
714 (The Windows -C command flag "wide API support" has been at least
715 temporarily retired in 5.8.1, and the -C has been repurposed, see
718 Most probably the right way to do this would be this:
719 L</"Virtualize operating system access">.
721 =head2 Unicode in %ENV
723 Currently the %ENV entries are always byte strings.
724 See L</"Virtualize operating system access">.
726 =head2 Unicode and glob()
728 Currently glob patterns and filenames returned from File::Glob::glob()
729 are always byte strings. See L</"Virtualize operating system access">.
731 =head2 use less 'memory'
733 Investigate trade offs to switch out perl's choices on memory usage.
734 Particularly perl should be able to give memory back.
736 This task is incremental - even a little bit of work on it will help.
738 =head2 Re-implement C<:unique> in a way that is actually thread-safe
740 The old implementation made bad assumptions on several levels. A good 90%
741 solution might be just to make C<:unique> work to share the string buffer
742 of SvPVs. That way large constant strings can be shared between ithreads,
743 such as the configuration information in F<Config>.
745 =head2 Make tainting consistent
747 Tainting would be easier to use if it didn't take documented shortcuts and
748 allow taint to "leak" everywhere within an expression.
750 =head2 readpipe(LIST)
752 system() accepts a LIST syntax (and a PROGRAM LIST syntax) to avoid
753 running a shell. readpipe() (the function behind qx//) could be similarly
756 =head2 Audit the code for destruction ordering assumptions
760 /* Need to check SvMAGICAL, as during global destruction it may be that
761 AvARYLEN(av) has been freed before av, and hence the SvANY() pointer
762 is now part of the linked list of SV heads, rather than pointing to
763 the original body. */
764 /* FIXME - audit the code for other bugs like this one. */
766 adding the C<SvMAGICAL> check to
768 if (AvARYLEN(av) && SvMAGICAL(AvARYLEN(av))) {
769 MAGIC *mg = mg_find (AvARYLEN(av), PERL_MAGIC_arylen);
771 Go through the core and look for similar assumptions that SVs have particular
772 types, as all bets are off during global destruction.
774 =head2 Extend PerlIO and PerlIO::Scalar
776 PerlIO::Scalar doesn't know how to truncate(). Implementing this
777 would require extending the PerlIO vtable.
779 Similarly the PerlIO vtable doesn't know about formats (write()), or
780 about stat(), or chmod()/chown(), utime(), or flock().
782 (For PerlIO::Scalar it's hard to see what e.g. mode bits or ownership
785 PerlIO doesn't do directories or symlinks, either: mkdir(), rmdir(),
786 opendir(), closedir(), seekdir(), rewinddir(), glob(); symlink(),
789 See also L</"Virtualize operating system access">.
791 =head2 Organize error messages
793 Perl's diagnostics (error messages, see L<perldiag>) could use
794 reorganizing and formalizing so that each error message has its
795 stable-for-all-eternity unique id, categorized by severity, type, and
796 subsystem. (The error messages would be listed in a datafile outside
797 of the Perl source code, and the source code would only refer to the
798 messages by the id.) This clean-up and regularizing should apply
799 for all croak() messages.
801 This would enable all sorts of things: easier translation/localization
802 of the messages (though please do keep in mind the caveats of
803 L<Locale::Maketext> about too straightforward approaches to
804 translation), filtering by severity, and instead of grepping for a
805 particular error message one could look for a stable error id. (Of
806 course, changing the error messages by default would break all the
807 existing software depending on some particular error message...)
809 This kind of functionality is known as I<message catalogs>. Look for
810 inspiration for example in the catgets() system, possibly even use it
811 if available-- but B<only> if available, all platforms will B<not>
814 For the really pure at heart, consider extending this item to cover
815 also the warning messages (see L<perllexwarn>, C<warnings.pl>).
817 =head1 Tasks that need a knowledge of the interpreter
819 These tasks would need C knowledge, and knowledge of how the interpreter works,
820 or a willingness to learn.
822 =head2 forbid labels with keyword names
824 Currently C<goto keyword> "computes" the label value:
826 $ perl -e 'goto print'
827 Can't find label 1 at -e line 1.
829 It is controversial if the right way to avoid the confusion is to forbid
830 labels with keyword names, or if it would be better to always treat
831 bareword expressions after a "goto" as a label and never as a keyword.
833 =head2 truncate() prototype
835 The prototype of truncate() is currently C<$$>. It should probably
836 be C<*$> instead. (This is changed in F<opcode.pl>)
838 =head2 error reporting of [$a ; $b]
840 Using C<;> inside brackets is a syntax error, and we don't propose to change
841 that by giving it any meaning. However, it's not reported very helpfully:
843 $ perl -e '$a = [$b; $c];'
844 syntax error at -e line 1, near "$b;"
845 syntax error at -e line 1, near "$c]"
846 Execution of -e aborted due to compilation errors.
848 It should be possible to hook into the tokeniser or the lexer, so that when a
849 C<;> is parsed where it is not legal as a statement terminator (ie inside
850 C<{}> used as a hashref, C<[]> or C<()>) it issues an error something like
851 I<';' isn't legal inside an expression - if you need multiple statements use a
852 do {...} block>. See the thread starting at
853 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-09/msg00573.html>
855 =head2 lexicals used only once
859 $ perl -we '$pie = 42'
860 Name "main::pie" used only once: possible typo at -e line 1.
864 $ perl -we 'my $pie = 42'
866 Logically all lexicals used only once should warn, if the user asks for
867 warnings. An unworked RT ticket (#5087) has been open for almost seven
868 years for this discrepancy.
872 The handling of Unicode is unclean in many places. In the regex engine
873 there are especially many problems. The swash data structure could be
874 replaced my something better. Inversion lists and maps are likely
875 candidates. The whole Unicode database could be placed in-core for a
876 huge speed-up. Only minimal work was done on the optimizer when utf8
877 was added, with the result that the synthetic start class often will
878 fail to narrow down the possible choices when given non-Latin1 input.
879 Karl Williamson has been working on this - talk to him.
881 =head2 state variable initialization in list context
883 Currently this is illegal:
885 state ($a, $b) = foo();
887 In Perl 6, C<state ($a) = foo();> and C<(state $a) = foo();> have different
888 semantics, which is tricky to implement in Perl 5 as currently they produce
889 the same opcode trees. The Perl 6 design is firm, so it would be good to
890 implement the necessary code in Perl 5. There are comments in
891 C<Perl_newASSIGNOP()> that show the code paths taken by various assignment
892 constructions involving state variables.
894 =head2 A does() built-in
896 Like ref(), only useful. It would call the C<DOES> method on objects; it
897 would also tell whether something can be dereferenced as an
898 array/hash/etc., or used as a regexp, etc.
899 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-03/msg00481.html>
901 =head2 Tied filehandles and write() don't mix
903 There is no method on tied filehandles to allow them to be called back by
906 =head2 Propagate compilation hints to the debugger
908 Currently a debugger started with -dE on the command-line doesn't see the
909 features enabled by -E. More generally hints (C<$^H> and C<%^H>) aren't
910 propagated to the debugger. Probably it would be a good thing to propagate
911 hints from the innermost non-C<DB::> scope: this would make code eval'ed
912 in the debugger see the features (and strictures, etc.) currently in
915 =head2 Attach/detach debugger from running program
917 The old perltodo notes "With C<gdb>, you can attach the debugger to a running
918 program if you pass the process ID. It would be good to do this with the Perl
919 debugger on a running Perl program, although I'm not sure how it would be
920 done." ssh and screen do this with named pipes in /tmp. Maybe we can too.
922 =head2 LVALUE functions for lists
924 The old perltodo notes that lvalue functions don't work for list or hash
925 slices. This would be good to fix.
927 =head2 regexp optimiser optional
929 The regexp optimiser is not optional. It should configurable to be, to allow
930 its performance to be measured, and its bugs to be easily demonstrated.
932 =head2 C</w> regex modifier
934 That flag would enable to match whole words, and also to interpolate
935 arrays as alternations. With it, C</P/w> would be roughly equivalent to:
937 do { local $"='|'; /\b(?:P)\b/ }
940 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-01/msg00400.html>
943 =head2 optional optimizer
945 Make the peephole optimizer optional. Currently it performs two tasks as
946 it walks the optree - genuine peephole optimisations, and necessary fixups of
947 ops. It would be good to find an efficient way to switch out the
948 optimisations whilst keeping the fixups.
950 =head2 You WANT *how* many
952 Currently contexts are void, scalar and list. split has a special mechanism in
953 place to pass in the number of return values wanted. It would be useful to
954 have a general mechanism for this, backwards compatible and little speed hit.
955 This would allow proposals such as short circuiting sort to be implemented
958 =head2 lexical aliases
960 Allow lexical aliases (maybe via the syntax C<my \$alias = \$foo>).
964 Self-ties are currently illegal because they caused too many segfaults. Maybe
965 the causes of these could be tracked down and self-ties on all types
968 =head2 Optimize away @_
970 The old perltodo notes "Look at the "reification" code in C<av.c>".
972 =head2 Virtualize operating system access
974 Implement a set of "vtables" that virtualizes operating system access
975 (open(), mkdir(), unlink(), readdir(), getenv(), etc.) At the very
976 least these interfaces should take SVs as "name" arguments instead of
977 bare char pointers; probably the most flexible and extensible way
978 would be for the Perl-facing interfaces to accept HVs. The system
979 needs to be per-operating-system and per-file-system
980 hookable/filterable, preferably both from XS and Perl level
981 (L<perlport/"Files and Filesystems"> is good reading at this point,
982 in fact, all of L<perlport> is.)
984 This has actually already been implemented (but only for Win32),
985 take a look at F<iperlsys.h> and F<win32/perlhost.h>. While all Win32
986 variants go through a set of "vtables" for operating system access,
987 non-Win32 systems currently go straight for the POSIX/Unix-style
988 system/library call. Similar system as for Win32 should be
989 implemented for all platforms. The existing Win32 implementation
990 probably does not need to survive alongside this proposed new
991 implementation, the approaches could be merged.
993 What would this give us? One often-asked-for feature this would
994 enable is using Unicode for filenames, and other "names" like %ENV,
995 usernames, hostnames, and so forth.
996 (See L<perlunicode/"When Unicode Does Not Happen">.)
998 But this kind of virtualization would also allow for things like
999 virtual filesystems, virtual networks, and "sandboxes" (though as long
1000 as dynamic loading of random object code is allowed, not very safe
1001 sandboxes since external code of course know not of Perl's vtables).
1002 An example of a smaller "sandbox" is that this feature can be used to
1003 implement per-thread working directories: Win32 already does this.
1005 See also L</"Extend PerlIO and PerlIO::Scalar">.
1007 =head2 Store the current pad in the OP slab allocator
1010 I hope that I got that "current pad" part correct
1012 Currently we leak ops in various cases of parse failure. I suggested that we
1013 could solve this by always using the op slab allocator, and walking it to
1014 free ops. Dave comments that as some ops are already freed during optree
1015 creation one would have to mark which ops are freed, and not double free them
1016 when walking the slab. He notes that one problem with this is that for some ops
1017 you have to know which pad was current at the time of allocation, which does
1018 change. I suggested storing a pointer to the current pad in the memory allocated
1019 for the slab, and swapping to a new slab each time the pad changes. Dave thinks
1020 that this would work.
1022 =head2 repack the optree
1024 Repacking the optree after execution order is determined could allow
1025 removal of NULL ops, and optimal ordering of OPs with respect to cache-line
1026 filling. The slab allocator could be reused for this purpose. I think that
1027 the best way to do this is to make it an optional step just before the
1028 completed optree is attached to anything else, and to use the slab allocator
1029 unchanged, so that freeing ops is identical whether or not this step runs.
1030 Note that the slab allocator allocates ops downwards in memory, so one would
1031 have to actually "allocate" the ops in reverse-execution order to get them
1032 contiguous in memory in execution order.
1035 L<http://www.nntp.perl.org/group/perl.perl5.porters/2007/12/msg131975.html>
1037 Note that running this copy, and then freeing all the old location ops would
1038 cause their slabs to be freed, which would eliminate possible memory wastage if
1039 the previous suggestion is implemented, and we swap slabs more frequently.
1041 =head2 eliminate incorrect line numbers in warnings
1049 } elsif ($undef == 0) {
1052 used to produce this output:
1054 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1055 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1057 where the line of the second warning was misreported - it should be line 5.
1058 Rafael fixed this - the problem arose because there was no nextstate OP
1059 between the execution of the C<if> and the C<elsif>, hence C<PL_curcop> still
1060 reports that the currently executing line is line 4. The solution was to inject
1061 a nextstate OPs for each C<elsif>, although it turned out that the nextstate
1062 OP needed to be a nulled OP, rather than a live nextstate OP, else other line
1063 numbers became misreported. (Jenga!)
1065 The problem is more general than C<elsif> (although the C<elsif> case is the
1066 most common and the most confusing). Ideally this code
1076 would produce this output
1078 Use of uninitialized value $undef in addition (+) at wrong.pl line 4.
1079 Use of uninitialized value $undef in addition (+) at wrong.pl line 7.
1081 (rather than lines 4 and 5), but this would seem to require every OP to carry
1082 (at least) line number information.
1084 What might work is to have an optional line number in memory just before the
1085 BASEOP structure, with a flag bit in the op to say whether it's present.
1086 Initially during compile every OP would carry its line number. Then add a late
1087 pass to the optimiser (potentially combined with L</repack the optree>) which
1088 looks at the two ops on every edge of the graph of the execution path. If
1089 the line number changes, flags the destination OP with this information.
1090 Once all paths are traced, replace every op with the flag with a
1091 nextstate-light op (that just updates C<PL_curcop>), which in turn then passes
1092 control on to the true op. All ops would then be replaced by variants that
1093 do not store the line number. (Which, logically, why it would work best in
1094 conjunction with L</repack the optree>, as that is already copying/reallocating
1097 (Although I should note that we're not certain that doing this for the general
1100 =head2 optimize tail-calls
1102 Tail-calls present an opportunity for broadly applicable optimization;
1103 anywhere that C<< return foo(...) >> is called, the outer return can
1104 be replaced by a goto, and foo will return directly to the outer
1105 caller, saving (conservatively) 25% of perl's call&return cost, which
1106 is relatively higher than in C. The scheme language is known to do
1107 this heavily. B::Concise provides good insight into where this
1108 optimization is possible, ie anywhere entersub,leavesub op-sequence
1111 perl -MO=Concise,-exec,a,b,-main -e 'sub a{ 1 }; sub b {a()}; b(2)'
1113 Bottom line on this is probably a new pp_tailcall function which
1114 combines the code in pp_entersub, pp_leavesub. This should probably
1115 be done 1st in XS, and using B::Generate to patch the new OP into the
1118 =head2 Add C<0odddd>
1120 It has been proposed that octal constants be specifiable through the syntax
1121 C<0oddddd>, parallel to the existing construct to specify hex constants
1126 Tasks that will get your name mentioned in the description of the "Highlights
1129 =head2 make ithreads more robust
1131 Generally make ithreads more robust.
1133 This task is incremental - even a little bit of work on it will help, and
1134 will be greatly appreciated.
1136 One bit would be to determine how to clone directory handles on systems
1137 without a C<fchdir> function (in sv.c:Perl_dirp_dup).
1139 Fix Perl_sv_dup, et al so that threads can return objects.
1141 =head2 Add class set operations to regexp engine
1143 Apparently these are quite useful. Anyway, Jeffery Friedl wants them.
1145 demerphq has this on his todo list, but right at the bottom.
1148 =head1 Tasks for microperl
1151 [ Each and every one of these may be obsolete, but they were listed
1152 in the old Todo.micro file]
1154 =head2 do away with fork/exec/wait?
1156 (system, popen should be enough?)
1158 =head2 some of the uconfig.sh really needs to be probed (using cc) in buildtime:
1160 (uConfigure? :-) native datatype widths and endianness come to mind