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 need only a little Perl knowledge
28 =head2 Fix POD errors in Perl documentation
30 Perl documentation is furnished in POD (Plain Old Documentation); see
31 L<perlpod>. We also have a utility that checks for various errors in
32 this documentation: F<t/porting/podcheck.t>. Unfortunately many files
33 have errors in them, and there is a database of known problems, kept in
34 F<t/porting/known_pod_issues.dat>. The most prevalent errors are lines
35 too wide to fit in a standard terminal window, but there are more
36 serious problems as well; and there are items listed there that are not
37 in fact errors. The task would be to go through and clean up the
38 documentation. This would be a good way to learn more about Perl.
40 =head1 Tasks that only need Perl knowledge
42 =head2 Classify bug tickets by type
44 Known bugs in Perl are tracked by L<https://rt.perl.org/> (which also
45 includes Perl 6). A summary can be found at
46 L<https://rt.perl.org/NoAuth/perl5/Overview.html>.
47 It shows bugs classified by "type". However, the type of many of the
48 bugs is "unknown". This greatly lowers the chances of them getting
49 fixed, as the number of open bugs is overwhelming -- too many to wade
50 through for someone to try to find the bugs in the parts of
51 Perl that s/he knows well enough to try to fix. This task involves
52 going through these bugs and classifying them into one or more types.
54 =head2 Ongoing: investigate new bug reports
56 When a bug report is filed, it would be very helpful to have someone do
57 a quick investigation to see if it is a real problem, and to reply to
58 the poster about it, asking for example code that reproduces the
59 problem. Such code should be added to the test suite as TODO tests, and
60 the ticket should be classified by type. To get started on this task,
61 look at the tickets that are marked as "New Issues" in
62 L<https://rt.perl.org/NoAuth/perl5/Overview.html>.
64 =head2 Migrate t/ from custom TAP generation
66 Many tests below F<t/> still generate TAP by "hand", rather than using library
67 functions. As explained in L<perlhack/TESTING>, tests in F<t/> are
68 written in a particular way to test that more complex constructions actually
69 work before using them routinely. Hence they don't use C<Test::More>, but
70 instead there is an intentionally simpler library, F<t/test.pl>. However,
71 quite a few tests in F<t/> have not been refactored to use it. Refactoring
72 any of these tests, one at a time, is a useful thing TODO.
74 The subdirectories F<base>, F<cmd> and F<comp>, that contain the most
75 basic tests, should be excluded from this task.
77 =head2 Automate perldelta generation
79 The perldelta file accompanying each release summaries the major changes.
80 It's mostly manually generated currently, but some of that could be
81 automated with a bit of perl, specifically the generation of
85 =item Modules and Pragmata
87 =item New Documentation
93 See F<Porting/how_to_write_a_perldelta.pod> for details.
95 =head2 Make Schwern poorer
97 We should have tests for everything. When all the core's modules are tested,
98 Schwern has promised to donate to $500 to TPF. We may need volunteers to
99 hold him upside down and shake vigorously in order to actually extract the
102 =head2 Write descriptions for all tests
104 Many individual tests in the test suite lack descriptions (or names, or labels
105 -- call them what you will). Many files completely lack descriptions, meaning
106 that the only output you get is the test numbers. If all tests had
107 descriptions, understanding what the tests are testing and why they sometimes
108 fail would both get a whole lot easier.
110 =head2 Improve the coverage of the core tests
112 Use Devel::Cover to ascertain the core modules' test coverage, then add
113 tests that are currently missing.
117 A full test suite for the B module would be nice.
119 =head2 A decent benchmark
121 C<perlbench> seems impervious to any recent changes made to the perl core. It
122 would be useful to have a reasonable general benchmarking suite that roughly
123 represented what current perl programs do, and measurably reported whether
124 tweaks to the core improve, degrade or don't really affect performance, to
125 guide people attempting to optimise the guts of perl. Gisle would welcome
126 new tests for perlbench. Steffen Schwingon would welcome help with
127 L<Benchmark::Perl::Formance>
129 =head2 fix tainting bugs
131 Fix the bugs revealed by running the test suite with the C<-t> switch.
132 Setting the TEST_ARGS environment variable to C<-taintwarn> will accomplish
135 =head2 Dual life everything
137 As part of the "dists" plan, anything that doesn't belong in the smallest perl
138 distribution needs to be dual lifed. Anything else can be too. Figure out what
139 changes would be needed to package that module and its tests up for CPAN, and
140 do so. Test it with older perl releases, and fix the problems you find.
142 To make a minimal perl distribution, it's useful to look at
143 F<t/lib/commonsense.t>.
145 =head2 POSIX memory footprint
147 Ilya observed that use POSIX; eats memory like there's no tomorrow, and at
148 various times worked to cut it down. There is probably still fat to cut out -
149 for example POSIX passes Exporter some very memory hungry data structures.
151 =head2 makedef.pl and conditional compilation
153 The script F<makedef.pl> that generates the list of exported symbols on
154 platforms which need this. Functions are declared in F<embed.fnc>, variables
155 in F<intrpvar.h>. Quite a few of the functions and variables are conditionally
156 declared there, using C<#ifdef>. However, F<makedef.pl> doesn't understand the
157 C macros, so the rules about which symbols are present when is duplicated in
158 the Perl code. Writing things twice is bad, m'kay. It would be good to teach
159 F<.pl> to understand the conditional compilation, and hence remove the
160 duplication, and the mistakes it has caused.
162 =head2 use strict; and AutoLoad
164 Currently if you write
167 use AutoLoader 'AUTOLOAD';
172 print join (' ', No, strict, here), "!\n";
175 then C<use strict;> isn't in force within the autoloaded subroutines. It would
176 be more consistent (and less surprising) to arrange for all lexical pragmas
177 in force at the __END__ block to be in force within each autoloaded subroutine.
179 There's a similar problem with SelfLoader.
181 =head2 profile installman
183 The F<installman> script is slow. All it is doing text processing, which we're
184 told is something Perl is good at. So it would be nice to know what it is doing
185 that is taking so much CPU, and where possible address it.
187 =head2 enable lexical enabling/disabling of individual warnings
189 Currently, warnings can only be enabled or disabled by category. There
190 are times when it would be useful to quash a single warning, not a
193 =head2 document diagnostics
195 Many diagnostic messages are not currently documented. The list is at the end
198 =head2 Write TODO tests for open bugs
200 Sometimes bugs get fixed as a side effect of something else, and
201 the bug remains open because no one realizes that it has been fixed.
202 Ideally, every open bug should have a TODO test in the core test suite.
204 =head1 Tasks that need a little sysadmin-type knowledge
206 Or if you prefer, tasks that you would learn from, and broaden your skills
209 =head2 make HTML install work
211 There is an C<install.html> target in the Makefile. It's marked as
212 "experimental". It would be good to get this tested, make it work reliably, and
213 remove the "experimental" tag. This would include
219 Checking that cross linking between various parts of the documentation works.
220 In particular that links work between the modules (files with POD in F<lib/>)
221 and the core documentation (files in F<pod/>)
225 Improving the code that split C<perlfunc> into chunks, preferably with
226 general case code added to L<Pod::Functions> that could be used elsewhere.
228 Challenges here are correctly identifying the groups of functions that go
229 together, and making the right named external cross-links point to the right
230 page. Currently this works reasonably well in the general case, and correctly
231 parses two or more C<=items> giving the different parameter lists for the
232 same function, such used by C<substr>. However it fails completely where
233 I<different> functions are listed as a sequence of C<=items> but share the
234 same description. All the functions from C<getpwnam> to C<endprotoent> have
235 individual stub pages, with only the page for C<endservent> holding the
236 description common to all. Likewise C<q>, C<qq> and C<qw> have stub pages,
237 instead of sharing the body of C<qx>.
239 Note also the current code isn't ideal with the two forms of C<select>, mushing
240 them both into one F<select.html> with the two descriptions run together.
241 Fixing this may well be a special case.
245 =head2 compressed man pages
247 Be able to install them. This would probably need a configure test to see how
248 the system does compressed man pages (same directory/different directory?
249 same filename/different filename), as well as tweaking the F<installman> script
250 to compress as necessary.
252 =head2 Add a code coverage target to the Makefile
254 Make it easy for anyone to run Devel::Cover on the core's tests. The steps
255 to do this manually are roughly
261 do a normal C<Configure>, but include Devel::Cover as a module to install
262 (see L<INSTALL> for how to do this)
270 cd t; HARNESS_PERL_SWITCHES=-MDevel::Cover ./perl -I../lib harness
274 Process the resulting Devel::Cover database
278 This just give you the coverage of the F<.pm>s. To also get the C level
285 Additionally tell C<Configure> to use the appropriate C compiler flags for
292 (instead of C<make perl>)
296 After running the tests run C<gcov> to generate all the F<.gcov> files.
297 (Including down in the subdirectories of F<ext/>
301 (From the top level perl directory) run C<gcov2perl> on all the C<.gcov> files
302 to get their stats into the cover_db directory.
306 Then process the Devel::Cover database
310 It would be good to add a single switch to C<Configure> to specify that you
311 wanted to perform perl level coverage, and another to specify C level
312 coverage, and have C<Configure> and the F<Makefile> do all the right things
315 =head2 Make Config.pm cope with differences between built and installed perl
317 Quite often vendors ship a perl binary compiled with their (pay-for)
318 compilers. People install a free compiler, such as gcc. To work out how to
319 build extensions, Perl interrogates C<%Config>, so in this situation
320 C<%Config> describes compilers that aren't there, and extension building
321 fails. This forces people into choosing between re-compiling perl themselves
322 using the compiler they have, or only using modules that the vendor ships.
324 It would be good to find a way teach C<Config.pm> about the installation setup,
325 possibly involving probing at install time or later, so that the C<%Config> in
326 a binary distribution better describes the installed machine, when the
327 installed machine differs from the build machine in some significant way.
329 =head2 linker specification files
331 Some platforms mandate that you provide a list of a shared library's external
332 symbols to the linker, so the core already has the infrastructure in place to
333 do this for generating shared perl libraries. Florian Ragwitz has been working
334 to offer this for the GNU toolchain, to allow Unix users to test that the
335 export list is correct, and to build a perl that does not pollute the global
336 namespace with private symbols, and will fail in the same way as msvc or mingw
337 builds or when using PERL_DL_NONLAZY=1. See the branch smoke-me/rafl/ld_export
339 =head2 Cross-compile support
341 We get requests for "how to cross compile Perl". The vast majority of these
342 seem to be for a couple of scenarios:
348 Platforms that could build natively using F<./Configure> (I<e.g.> Linux or
349 NetBSD on MIPS or ARM) but people want to use a beefier machine (and on the
350 same OS) to build more easily.
354 Platforms that can't build natively, but no (significant) porting changes
355 are needed to our current source code. Prime example of this is Android.
359 There are several scripts and tools for cross-compiling perl for other
360 platforms. However, these are somewhat inconsistent and scattered across the
361 codebase, none are documented well, none are clearly flexible enough to
362 be confident that they can support any TARGET/HOST platform pair other than
363 that which they were developed on, and it's not clear how bitrotted they are.
365 For example, C<Configure> understands C<-Dusecrosscompile> option. This option
366 arranges for building C<miniperl> for TARGET machine, so this C<miniperl> is
367 assumed then to be copied to TARGET machine and used as a replacement of
368 full C<perl> executable. This code is almost 10 years old. Meanwhile, the
369 F<Cross/> directory contains two different approaches for cross compiling to
370 ARM Linux targets, relying on hand curated F<config.sh> files, but that code
371 is getting on for 5 years old, and requires insider knowledge of perl's
372 build system to draft a F<config.sh> for a new platform.
374 Jess Robinson has submitted a grant to TPF to work on cleaning this up.
376 =head2 Split "linker" from "compiler"
378 Right now, Configure probes for two commands, and sets two variables:
382 =item * C<cc> (in F<cc.U>)
384 This variable holds the name of a command to execute a C compiler which
385 can resolve multiple global references that happen to have the same
386 name. Usual values are F<cc> and F<gcc>.
387 Fervent ANSI compilers may be called F<c89>. AIX has F<xlc>.
389 =item * C<ld> (in F<dlsrc.U>)
391 This variable indicates the program to be used to link
392 libraries for dynamic loading. On some systems, it is F<ld>.
393 On ELF systems, it should be C<$cc>. Mostly, we'll try to respect
394 the hint file setting.
398 There is an implicit historical assumption from around Perl5.000alpha
399 something, that C<$cc> is also the correct command for linking object files
400 together to make an executable. This may be true on Unix, but it's not true
401 on other platforms, and there are a maze of work arounds in other places (such
402 as F<Makefile.SH>) to cope with this.
404 Ideally, we should create a new variable to hold the name of the executable
405 linker program, probe for it in F<Configure>, and centralise all the special
406 case logic there or in hints files.
408 A small bikeshed issue remains - what to call it, given that C<$ld> is already
409 taken (arguably for the wrong thing now, but on SunOS 4.1 it is the command
410 for creating dynamically-loadable modules) and C<$link> could be confused with
411 the Unix command line executable of the same name, which does something
412 completely different. Andy Dougherty makes the counter argument "In parrot, I
413 tried to call the command used to link object files and libraries into an
414 executable F<link>, since that's what my vaguely-remembered DOS and VMS
415 experience suggested. I don't think any real confusion has ensued, so it's
416 probably a reasonable name for perl5 to use."
418 "Alas, I've always worried that introducing it would make things worse,
419 since now the module building utilities would have to look for
420 C<$Config{link}> and institute a fall-back plan if it weren't found."
421 Although I can see that as confusing, given that C<$Config{d_link}> is true
422 when (hard) links are available.
424 =head2 Configure Windows using PowerShell
426 Currently, Windows uses hard-coded config files based to build the
427 config.h for compiling Perl. Makefiles are also hard-coded and need to be
428 hand edited prior to building Perl. While this makes it easy to create a perl.exe
429 that works across multiple Windows versions, being able to accurately
430 configure a perl.exe for a specific Windows versions and VS C++ would be
431 a nice enhancement. With PowerShell available on Windows XP and up, this
432 may now be possible. Step 1 might be to investigate whether this is possible
433 and use this to clean up our current makefile situation. Step 2 would be to
434 see if there would be a way to use our existing metaconfig units to configure a
435 Windows Perl or whether we go in a separate direction and make it so. Of
436 course, we all know what step 3 is.
438 =head1 Tasks that need a little C knowledge
440 These tasks would need a little C knowledge, but don't need any specific
441 background or experience with XS, or how the Perl interpreter works
443 =head2 Weed out needless PERL_UNUSED_ARG
445 The C code uses the macro C<PERL_UNUSED_ARG> to stop compilers warning about
446 unused arguments. Often the arguments can't be removed, as there is an
447 external constraint that determines the prototype of the function, so this
448 approach is valid. However, there are some cases where C<PERL_UNUSED_ARG>
449 could be removed. Specifically
455 The prototypes of (nearly all) static functions can be changed
459 Unused arguments generated by short cut macros are wasteful - the short cut
460 macro used can be changed.
466 Natively 64-bit systems need neither -Duse64bitint nor -Duse64bitall.
467 On these systems, it might be the default compilation mode, and there
468 is currently no guarantee that passing no use64bitall option to the
469 Configure process will build a 32bit perl. Implementing -Duse32bit*
470 options would be nice for perl 5.19.11.
472 =head2 Profile Perl - am I hot or not?
474 The Perl source code is stable enough that it makes sense to profile it,
475 identify and optimise the hotspots. It would be good to measure the
476 performance of the Perl interpreter using free tools such as cachegrind,
477 gprof, and dtrace, and work to reduce the bottlenecks they reveal.
479 As part of this, the idea of F<pp_hot.c> is that it contains the I<hot> ops,
480 the ops that are most commonly used. The idea is that by grouping them, their
481 object code will be adjacent in the executable, so they have a greater chance
482 of already being in the CPU cache (or swapped in) due to being near another op
485 Except that it's not clear if these really are the most commonly used ops. So
486 as part of exercising your skills with coverage and profiling tools you might
487 want to determine what ops I<really> are the most commonly used. And in turn
488 suggest evictions and promotions to achieve a better F<pp_hot.c>.
490 One piece of Perl code that might make a good testbed is F<installman>.
492 =head2 Improve win32/wince.c
494 Currently, numerous functions look virtually, if not completely,
495 identical in both F<win32/wince.c> and F<win32/win32.c> files, which can't
498 =head2 Use secure CRT functions when building with VC8 on Win32
500 Visual C++ 2005 (VC++ 8.x) deprecated a number of CRT functions on the basis
501 that they were "unsafe" and introduced differently named secure versions of
502 them as replacements, e.g. instead of writing
504 FILE* f = fopen(__FILE__, "r");
509 errno_t err = fopen_s(&f, __FILE__, "r");
511 Currently, the warnings about these deprecations have been disabled by adding
512 -D_CRT_SECURE_NO_DEPRECATE to the CFLAGS. It would be nice to remove that
513 warning suppressant and actually make use of the new secure CRT functions.
515 There is also a similar issue with POSIX CRT function names like fileno having
516 been deprecated in favour of ISO C++ conformant names like _fileno. These
517 warnings are also currently suppressed by adding -D_CRT_NONSTDC_NO_DEPRECATE. It
518 might be nice to do as Microsoft suggest here too, although, unlike the secure
519 functions issue, there is presumably little or no benefit in this case.
521 =head2 Fix POSIX::access() and chdir() on Win32
523 These functions currently take no account of DACLs and therefore do not behave
524 correctly in situations where access is restricted by DACLs (as opposed to the
525 read-only attribute).
527 Furthermore, POSIX::access() behaves differently for directories having the
528 read-only attribute set depending on what CRT library is being used. For
529 example, the _access() function in the VC6 and VC7 CRTs (wrongly) claim that
530 such directories are not writable, whereas in fact all directories are writable
531 unless access is denied by DACLs. (In the case of directories, the read-only
532 attribute actually only means that the directory cannot be deleted.) This CRT
533 bug is fixed in the VC8 and VC9 CRTs (but, of course, the directory may still
534 not actually be writable if access is indeed denied by DACLs).
536 For the chdir() issue, see ActiveState bug #74552:
537 L<http://bugs.activestate.com/show_bug.cgi?id=74552>
539 Therefore, DACLs should be checked both for consistency across CRTs and for
542 (Note that perl's -w operator should not be modified to check DACLs. It has
543 been written so that it reflects the state of the read-only attribute, even
544 for directories (whatever CRT is being used), for symmetry with chmod().)
546 =head2 strcat(), strcpy(), strncat(), strncpy(), sprintf(), vsprintf()
548 Maybe create a utility that checks after each libperl.a creation that
549 none of the above (nor sprintf(), vsprintf(), or *SHUDDER* gets())
550 ever creep back to libperl.a.
552 nm libperl.a | ./miniperl -alne '$o = $F[0] if /:$/;
553 print "$o $F[1]" if $F[0] eq "U" && $F[1] =~ /^(?:strn?c(?:at|py)|v?sprintf|gets)$/'
555 Note, of course, that this will only tell whether B<your> platform
556 is using those naughty interfaces.
558 =head2 -D_FORTIFY_SOURCE=2
560 Recent glibcs support C<-D_FORTIFY_SOURCE=2> which gives
561 protection against various kinds of buffer overflow problems.
562 It should probably be used for compiling Perl whenever available,
563 Configure and/or hints files should be adjusted to probe for the
564 availability of these feature and enable it as appropriate.
566 =head2 Arenas for GPs? For MAGIC?
568 C<struct gp> and C<struct magic> are both currently allocated by C<malloc>.
569 It might be a speed or memory saving to change to using arenas. Or it might
570 not. It would need some suitable benchmarking first. In particular, C<GP>s
571 can probably be changed with minimal compatibility impact (probably nothing
572 outside of the core, or even outside of F<gv.c> allocates them), but they
573 probably aren't allocated/deallocated often enough for a speed saving. Whereas
574 C<MAGIC> is allocated/deallocated more often, but in turn, is also something
575 more externally visible, so changing the rules here may bite external code.
579 Several SV body structs are now the same size, notably PVMG and PVGV, PVAV and
580 PVHV, and PVCV and PVFM. It should be possible to allocate and return same
581 sized bodies from the same actual arena, rather than maintaining one arena for
582 each. This could save 4-6K per thread, of memory no longer tied up in the
583 not-yet-allocated part of an arena.
586 =head1 Tasks that need a knowledge of XS
588 These tasks would need C knowledge, and roughly the level of knowledge of
589 the perl API that comes from writing modules that use XS to interface to
592 =head2 Write an XS cookbook
594 Create pod/perlxscookbook.pod with short, task-focused 'recipes' in XS that
595 demonstrate common tasks and good practices. (Some of these might be
596 extracted from perlguts.) The target audience should be XS novices, who need
597 more examples than perlguts but something less overwhelming than perlapi.
598 Recipes should provide "one pretty good way to do it" instead of TIMTOWTDI.
600 Rather than focusing on interfacing Perl to C libraries, such a cookbook
601 should probably focus on how to optimize Perl routines by re-writing them
602 in XS. This will likely be more motivating to those who mostly work in
603 Perl but are looking to take the next step into XS.
605 Deconstructing and explaining some simpler XS modules could be one way to
606 bootstrap a cookbook. (List::Util? Class::XSAccessor? Tree::Ternary_XS?)
607 Another option could be deconstructing the implementation of some simpler
610 =head2 Document how XSUBs can use C<cv_set_call_checker> to inline themselves as OPs
612 For a simple XSUB, often the subroutine dispatch takes more time than the
613 XSUB itself. v5.14.0 now allows XSUBs to register a function which will be
614 called when the parser is finished building an C<entersub> op which calls
617 Registration is done with C<Perl_cv_set_call_checker>, is documented at the
618 API level in L<perlapi>, and L<perl5140delta/Custom per-subroutine check hooks>
619 notes that it can be used to inline a subroutine, by replacing it with a
620 custom op. However there is no further detail of the code needed to do this.
621 It would be useful to add one or more annotated examples of how to create
624 This should provide a measurable speed up to simple XSUBs inside
625 tight loops. Initially one would have to write the OP alternative
626 implementation by hand, but it's likely that this should be reasonably
627 straightforward for the type of XSUB that would benefit the most. Longer
628 term, once the run-time implementation is proven, it should be possible to
629 progressively update ExtUtils::ParseXS to generate OP implementations for
632 =head2 Remove the use of SVs as temporaries in dump.c
634 F<dump.c> contains debugging routines to dump out the contains of perl data
635 structures, such as C<SV>s, C<AV>s and C<HV>s. Currently, the dumping code
636 B<uses> C<SV>s for its temporary buffers, which was a logical initial
637 implementation choice, as they provide ready made memory handling.
639 However, they also lead to a lot of confusion when it happens that what you're
640 trying to debug is seen by the code in F<dump.c>, correctly or incorrectly, as
641 a temporary scalar it can use for a temporary buffer. It's also not possible
642 to dump scalars before the interpreter is properly set up, such as during
643 ithreads cloning. It would be good to progressively replace the use of scalars
644 as string accumulation buffers with something much simpler, directly allocated
645 by C<malloc>. The F<dump.c> code is (or should be) only producing 7 bit
646 US-ASCII, so output character sets are not an issue.
648 Producing and proving an internal simple buffer allocation would make it easier
649 to re-write the internals of the PerlIO subsystem to avoid using C<SV>s for
650 B<its> buffers, use of which can cause problems similar to those of F<dump.c>,
653 =head2 safely supporting POSIX SA_SIGINFO
655 Some years ago Jarkko supplied patches to provide support for the POSIX
656 SA_SIGINFO feature in Perl, passing the extra data to the Perl signal handler.
658 Unfortunately, it only works with "unsafe" signals, because under safe
659 signals, by the time Perl gets to run the signal handler, the extra
660 information has been lost. Moreover, it's not easy to store it somewhere,
661 as you can't call mutexs, or do anything else fancy, from inside a signal
664 So it strikes me that we could provide safe SA_SIGINFO support
670 Provide global variables for two file descriptors
674 When the first request is made via C<sigaction> for C<SA_SIGINFO>, create a
675 pipe, store the reader in one, the writer in the other
679 In the "safe" signal handler (C<Perl_csighandler()>/C<S_raise_signal()>), if
680 the C<siginfo_t> pointer non-C<NULL>, and the writer file handle is open,
686 serialise signal number, C<struct siginfo_t> (or at least the parts we care
687 about) into a small auto char buff
691 C<write()> that (non-blocking) to the writer fd
697 if it writes 100%, flag the signal in a counter of "signals on the pipe" akin
698 to the current per-signal-number counts
702 if it writes 0%, assume the pipe is full. Flag the data as lost?
706 if it writes partially, croak a panic, as your OS is broken.
714 in the regular C<PERL_ASYNC_CHECK()> processing, if there are "signals on
715 the pipe", read the data out, deserialise, build the Perl structures on
716 the stack (code in C<Perl_sighandler()>, the "unsafe" handler), and call as
721 I think that this gets us decent C<SA_SIGINFO> support, without the current risk
722 of running Perl code inside the signal handler context. (With all the dangers
723 of things like C<malloc> corruption that that currently offers us)
725 For more information see the thread starting with this message:
726 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-03/msg00305.html>
728 =head2 autovivification
730 Make all autovivification consistent w.r.t LVALUE/RVALUE and strict/no strict;
732 This task is incremental - even a little bit of work on it will help.
734 =head2 Unicode in Filenames
736 chdir, chmod, chown, chroot, exec, glob, link, lstat, mkdir, open,
737 opendir, qx, readdir, readlink, rename, rmdir, stat, symlink, sysopen,
738 system, truncate, unlink, utime, -X. All these could potentially accept
739 Unicode filenames either as input or output (and in the case of system
740 and qx Unicode in general, as input or output to/from the shell).
741 Whether a filesystem - an operating system pair understands Unicode in
744 Known combinations that have some level of understanding include
745 Microsoft NTFS, Apple HFS+ (In Mac OS 9 and X) and Apple UFS (in Mac
746 OS X), NFS v4 is rumored to be Unicode, and of course Plan 9. How to
747 create Unicode filenames, what forms of Unicode are accepted and used
748 (UCS-2, UTF-16, UTF-8), what (if any) is the normalization form used,
749 and so on, varies. Finding the right level of interfacing to Perl
750 requires some thought. Remember that an OS does not implicate a
753 (The Windows -C command flag "wide API support" has been at least
754 temporarily retired in 5.8.1, and the -C has been repurposed, see
757 Most probably the right way to do this would be this:
758 L</"Virtualize operating system access">.
760 =head2 Unicode in %ENV
762 Currently the %ENV entries are always byte strings.
763 See L</"Virtualize operating system access">.
765 (See RT ticket #113536 for information on Win32's handling of %ENV,
766 which was fixed to work with native ANSI codepage characters in the
767 environment, but still doesn't work with other characters outside of
768 that codepage present in the environment.)
770 =head2 Unicode and glob()
772 Currently glob patterns and filenames returned from File::Glob::glob()
773 are always byte strings. See L</"Virtualize operating system access">.
775 =head2 use less 'memory'
777 Investigate trade offs to switch out perl's choices on memory usage.
778 Particularly perl should be able to give memory back.
780 This task is incremental - even a little bit of work on it will help.
782 =head2 Re-implement C<:unique> in a way that is actually thread-safe
784 The old implementation made bad assumptions on several levels. A good 90%
785 solution might be just to make C<:unique> work to share the string buffer
786 of SvPVs. That way large constant strings can be shared between ithreads,
787 such as the configuration information in F<Config>.
789 =head2 Make tainting consistent
791 Tainting would be easier to use if it didn't take documented shortcuts and
792 allow taint to "leak" everywhere within an expression.
794 =head2 readpipe(LIST)
796 system() accepts a LIST syntax (and a PROGRAM LIST syntax) to avoid
797 running a shell. readpipe() (the function behind qx//) could be similarly
800 =head2 Audit the code for destruction ordering assumptions
804 /* Need to check SvMAGICAL, as during global destruction it may be that
805 AvARYLEN(av) has been freed before av, and hence the SvANY() pointer
806 is now part of the linked list of SV heads, rather than pointing to
807 the original body. */
808 /* FIXME - audit the code for other bugs like this one. */
810 adding the C<SvMAGICAL> check to
812 if (AvARYLEN(av) && SvMAGICAL(AvARYLEN(av))) {
813 MAGIC *mg = mg_find (AvARYLEN(av), PERL_MAGIC_arylen);
815 Go through the core and look for similar assumptions that SVs have particular
816 types, as all bets are off during global destruction.
818 =head2 Extend PerlIO and PerlIO::Scalar
820 PerlIO::Scalar doesn't know how to truncate(). Implementing this
821 would require extending the PerlIO vtable.
823 Similarly the PerlIO vtable doesn't know about formats (write()), or
824 about stat(), or chmod()/chown(), utime(), or flock().
826 (For PerlIO::Scalar it's hard to see what e.g. mode bits or ownership
829 PerlIO doesn't do directories or symlinks, either: mkdir(), rmdir(),
830 opendir(), closedir(), seekdir(), rewinddir(), glob(); symlink(),
833 See also L</"Virtualize operating system access">.
835 =head2 Organize error messages
837 Perl's diagnostics (error messages, see L<perldiag>) could use
838 reorganizing and formalizing so that each error message has its
839 stable-for-all-eternity unique id, categorized by severity, type, and
840 subsystem. (The error messages would be listed in a datafile outside
841 of the Perl source code, and the source code would only refer to the
842 messages by the id.) This clean-up and regularizing should apply
843 for all croak() messages.
845 This would enable all sorts of things: easier translation/localization
846 of the messages (though please do keep in mind the caveats of
847 L<Locale::Maketext> about too straightforward approaches to
848 translation), filtering by severity, and instead of grepping for a
849 particular error message one could look for a stable error id. (Of
850 course, changing the error messages by default would break all the
851 existing software depending on some particular error message...)
853 This kind of functionality is known as I<message catalogs>. Look for
854 inspiration for example in the catgets() system, possibly even use it
855 if available-- but B<only> if available, all platforms will B<not>
858 For the really pure at heart, consider extending this item to cover
859 also the warning messages (see L<warnings>, F<regen/warnings.pl>).
861 =head1 Tasks that need a knowledge of the interpreter
863 These tasks would need C knowledge, and knowledge of how the interpreter works,
864 or a willingness to learn.
866 =head2 forbid labels with keyword names
868 Currently C<goto keyword> "computes" the label value:
870 $ perl -e 'goto print'
871 Can't find label 1 at -e line 1.
873 It is controversial if the right way to avoid the confusion is to forbid
874 labels with keyword names, or if it would be better to always treat
875 bareword expressions after a "goto" as a label and never as a keyword.
877 =head2 truncate() prototype
879 The prototype of truncate() is currently C<$$>. It should probably
880 be C<*$> instead. (This is changed in F<opcode.pl>)
882 =head2 error reporting of [$a ; $b]
884 Using C<;> inside brackets is a syntax error, and we don't propose to change
885 that by giving it any meaning. However, it's not reported very helpfully:
887 $ perl -e '$a = [$b; $c];'
888 syntax error at -e line 1, near "$b;"
889 syntax error at -e line 1, near "$c]"
890 Execution of -e aborted due to compilation errors.
892 It should be possible to hook into the tokeniser or the lexer, so that when a
893 C<;> is parsed where it is not legal as a statement terminator (ie inside
894 C<{}> used as a hashref, C<[]> or C<()>) it issues an error something like
895 I<';' isn't legal inside an expression - if you need multiple statements use a
896 do {...} block>. See the thread starting at
897 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-09/msg00573.html>
899 =head2 strict as warnings
901 See L<http://markmail.org/message/vbrupaslr3bybmvk>, where Josua ben Jore
902 writes: I've been of the opinion that everything strict.pm does ought to be
903 able to considered just warnings that have been promoted to 'FATAL'.
905 =head2 lexicals used only once
909 $ perl -we '$pie = 42'
910 Name "main::pie" used only once: possible typo at -e line 1.
914 $ perl -we 'my $pie = 42'
916 Logically all lexicals used only once should warn, if the user asks for
917 warnings. An unworked RT ticket (#5087) has been open for almost seven
918 years for this discrepancy.
922 The handling of Unicode is unclean in many places. In the regex engine
923 there are especially many problems. The swash data structure could be
924 replaced my something better. Inversion lists and maps are likely
925 candidates. The whole Unicode database could be placed in-core for a
926 huge speed-up. Only minimal work was done on the optimizer when utf8
927 was added, with the result that the synthetic start class often will
928 fail to narrow down the possible choices when given non-Latin1 input.
929 Karl Williamson has been working on this - talk to him.
931 =head2 state variable initialization in list context
933 Currently this is illegal:
935 state ($a, $b) = foo();
937 In Perl 6, C<state ($a) = foo();> and C<(state $a) = foo();> have different
938 semantics, which is tricky to implement in Perl 5 as currently they produce
939 the same opcode trees. The Perl 6 design is firm, so it would be good to
940 implement the necessary code in Perl 5. There are comments in
941 C<Perl_newASSIGNOP()> that show the code paths taken by various assignment
942 constructions involving state variables.
944 =head2 A does() built-in
946 Like ref(), only useful. It would call the C<DOES> method on objects; it
947 would also tell whether something can be dereferenced as an
948 array/hash/etc., or used as a regexp, etc.
949 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-03/msg00481.html>
951 =head2 Tied filehandles and write() don't mix
953 There is no method on tied filehandles to allow them to be called back by
956 =head2 Propagate compilation hints to the debugger
958 Currently a debugger started with -dE on the command-line doesn't see the
959 features enabled by -E. More generally hints (C<$^H> and C<%^H>) aren't
960 propagated to the debugger. Probably it would be a good thing to propagate
961 hints from the innermost non-C<DB::> scope: this would make code eval'ed
962 in the debugger see the features (and strictures, etc.) currently in
965 =head2 Attach/detach debugger from running program
967 The old perltodo notes "With C<gdb>, you can attach the debugger to a running
968 program if you pass the process ID. It would be good to do this with the Perl
969 debugger on a running Perl program, although I'm not sure how it would be
970 done." ssh and screen do this with named pipes in /tmp. Maybe we can too.
972 =head2 LVALUE functions for lists
974 The old perltodo notes that lvalue functions don't work for list or hash
975 slices. This would be good to fix.
977 =head2 regexp optimizer optional
979 The regexp optimizer is not optional. It should be configurable to be optional
980 and to allow its performance to be measured and its bugs to be easily
983 =head2 C</w> regex modifier
985 That flag would enable to match whole words, and also to interpolate
986 arrays as alternations. With it, C</P/w> would be roughly equivalent to:
988 do { local $"='|'; /\b(?:P)\b/ }
991 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-01/msg00400.html>
994 =head2 optional optimizer
996 Make the peephole optimizer optional. Currently it performs two tasks as
997 it walks the optree - genuine peephole optimisations, and necessary fixups of
998 ops. It would be good to find an efficient way to switch out the
999 optimisations whilst keeping the fixups.
1001 =head2 You WANT *how* many
1003 Currently contexts are void, scalar and list. split has a special mechanism in
1004 place to pass in the number of return values wanted. It would be useful to
1005 have a general mechanism for this, backwards compatible and little speed hit.
1006 This would allow proposals such as short circuiting sort to be implemented
1007 as a module on CPAN.
1009 =head2 lexical aliases
1011 Allow lexical aliases (maybe via the syntax C<my \$alias = \$foo>).
1015 Self-ties are currently illegal because they caused too many segfaults. Maybe
1016 the causes of these could be tracked down and self-ties on all types
1019 =head2 Optimize away @_
1021 The old perltodo notes "Look at the "reification" code in C<av.c>".
1023 =head2 Virtualize operating system access
1025 Implement a set of "vtables" that virtualizes operating system access
1026 (open(), mkdir(), unlink(), readdir(), getenv(), etc.) At the very
1027 least these interfaces should take SVs as "name" arguments instead of
1028 bare char pointers; probably the most flexible and extensible way
1029 would be for the Perl-facing interfaces to accept HVs. The system
1030 needs to be per-operating-system and per-file-system
1031 hookable/filterable, preferably both from XS and Perl level
1032 (L<perlport/"Files and Filesystems"> is good reading at this point,
1033 in fact, all of L<perlport> is.)
1035 This has actually already been implemented (but only for Win32),
1036 take a look at F<iperlsys.h> and F<win32/perlhost.h>. While all Win32
1037 variants go through a set of "vtables" for operating system access,
1038 non-Win32 systems currently go straight for the POSIX/Unix-style
1039 system/library call. Similar system as for Win32 should be
1040 implemented for all platforms. The existing Win32 implementation
1041 probably does not need to survive alongside this proposed new
1042 implementation, the approaches could be merged.
1044 What would this give us? One often-asked-for feature this would
1045 enable is using Unicode for filenames, and other "names" like %ENV,
1046 usernames, hostnames, and so forth.
1047 (See L<perlunicode/"When Unicode Does Not Happen">.)
1049 But this kind of virtualization would also allow for things like
1050 virtual filesystems, virtual networks, and "sandboxes" (though as long
1051 as dynamic loading of random object code is allowed, not very safe
1052 sandboxes since external code of course know not of Perl's vtables).
1053 An example of a smaller "sandbox" is that this feature can be used to
1054 implement per-thread working directories: Win32 already does this.
1056 See also L</"Extend PerlIO and PerlIO::Scalar">.
1058 =head2 repack the optree
1060 Repacking the optree after execution order is determined could allow
1061 removal of NULL ops, and optimal ordering of OPs with respect to cache-line
1062 filling. I think that
1063 the best way to do this is to make it an optional step just before the
1064 completed optree is attached to anything else, and to use the slab allocator
1065 unchanged--but allocate a single slab the right size, avoiding partial
1066 slabs--, so that freeing ops is identical whether or not this step runs.
1067 Note that the slab allocator allocates ops downwards in memory, so one would
1068 have to actually "allocate" the ops in reverse-execution order to get them
1069 contiguous in memory in execution order.
1072 L<http://www.nntp.perl.org/group/perl.perl5.porters/2007/12/msg131975.html>
1074 Note that running this copy, and then freeing all the old location ops would
1075 cause their slabs to be freed, which would eliminate possible memory wastage if
1076 the previous suggestion is implemented, and we swap slabs more frequently.
1078 =head2 eliminate incorrect line numbers in warnings
1086 } elsif ($undef == 0) {
1089 used to produce this output:
1091 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1092 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1094 where the line of the second warning was misreported - it should be line 5.
1095 Rafael fixed this - the problem arose because there was no nextstate OP
1096 between the execution of the C<if> and the C<elsif>, hence C<PL_curcop> still
1097 reports that the currently executing line is line 4. The solution was to inject
1098 a nextstate OPs for each C<elsif>, although it turned out that the nextstate
1099 OP needed to be a nulled OP, rather than a live nextstate OP, else other line
1100 numbers became misreported. (Jenga!)
1102 The problem is more general than C<elsif> (although the C<elsif> case is the
1103 most common and the most confusing). Ideally this code
1113 would produce this output
1115 Use of uninitialized value $undef in addition (+) at wrong.pl line 4.
1116 Use of uninitialized value $undef in addition (+) at wrong.pl line 7.
1118 (rather than lines 4 and 5), but this would seem to require every OP to carry
1119 (at least) line number information.
1121 What might work is to have an optional line number in memory just before the
1122 BASEOP structure, with a flag bit in the op to say whether it's present.
1123 Initially during compile every OP would carry its line number. Then add a late
1124 pass to the optimizer (potentially combined with L</repack the optree>) which
1125 looks at the two ops on every edge of the graph of the execution path. If
1126 the line number changes, flags the destination OP with this information.
1127 Once all paths are traced, replace every op with the flag with a
1128 nextstate-light op (that just updates C<PL_curcop>), which in turn then passes
1129 control on to the true op. All ops would then be replaced by variants that
1130 do not store the line number. (Which, logically, why it would work best in
1131 conjunction with L</repack the optree>, as that is already copying/reallocating
1134 (Although I should note that we're not certain that doing this for the general
1137 =head2 optimize tail-calls
1139 Tail-calls present an opportunity for broadly applicable optimization;
1140 anywhere that C<< return foo(...) >> is called, the outer return can
1141 be replaced by a goto, and foo will return directly to the outer
1142 caller, saving (conservatively) 25% of perl's call&return cost, which
1143 is relatively higher than in C. The scheme language is known to do
1144 this heavily. B::Concise provides good insight into where this
1145 optimization is possible, ie anywhere entersub,leavesub op-sequence
1148 perl -MO=Concise,-exec,a,b,-main -e 'sub a{ 1 }; sub b {a()}; b(2)'
1150 Bottom line on this is probably a new pp_tailcall function which
1151 combines the code in pp_entersub, pp_leavesub. This should probably
1152 be done 1st in XS, and using B::Generate to patch the new OP into the
1155 =head2 Add C<0odddd>
1157 It has been proposed that octal constants be specifiable through the syntax
1158 C<0oddddd>, parallel to the existing construct to specify hex constants
1161 =head2 Revisit the regex super-linear cache code
1163 Perl executes regexes using the traditional backtracking algorithm, which
1164 makes it possible to implement a variety of powerful pattern-matching
1165 features (like embedded code blocks), at the cost of taking exponential time
1166 to run on some pathological patterns. The exponential-time problem is
1167 mitigated by the I<super-linear cache>, which detects when we're processing
1168 such a pathological pattern, and does some additional bookkeeping to avoid
1169 much of the work. However, that code has bit-rotted a little; some patterns
1170 don't make as much use of it as they should. The proposal is to analyse
1171 where the current cache code has problems, and extend it to cover those cases.
1174 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2013-01/msg00339.html>
1178 Tasks that will get your name mentioned in the description of the "Highlights
1181 =head2 make ithreads more robust
1183 Generally make ithreads more robust.
1185 This task is incremental - even a little bit of work on it will help, and
1186 will be greatly appreciated.
1188 One bit would be to determine how to clone directory handles on systems
1189 without a C<fchdir> function (in sv.c:Perl_dirp_dup).
1191 Fix Perl_sv_dup, et al so that threads can return objects.
1193 =head1 Tasks for microperl
1196 [ Each and every one of these may be obsolete, but they were listed
1197 in the old Todo.micro file]
1199 =head2 do away with fork/exec/wait?
1201 (system, popen should be enough?)
1203 =head2 some of the uconfig.sh really needs to be probed (using cc) in buildtime:
1205 (uConfigure? :-) native datatype widths and endianness come to mind