3 perltodo - Perl TO-DO List
7 This is a list of wishes for Perl. The tasks we think are smaller or
8 easier are listed first. Anyone is welcome to work on any of these,
9 but it's a good idea to first contact I<perl5-porters@perl.org> to
10 avoid duplication of effort, and to learn from any previous attempts.
11 By all means contact a pumpking privately first if you prefer.
13 Whilst patches to make the list shorter are most welcome, ideas to add to
14 the list are also encouraged. Check the perl5-porters archives for past
15 ideas, and any discussion about them. One set of archives may be found at:
17 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/
19 What can we offer you in return? Fame, fortune, and everlasting glory? Maybe
20 not, but if your patch is incorporated, then we'll add your name to the
21 F<AUTHORS> file, which ships in the official distribution. How many other
22 programming languages offer you 1 line of immortality?
24 =head1 Tasks that only need Perl knowledge
26 =head2 Smartmatch design issues
28 In 5.10.0 the smartmatch operator C<~~> isn't working quite "right". But
29 before we can fix the implementation, we need to define what "right" is.
30 The first problem is that Robin Houston implemented the Perl 6 smart match
31 spec as of February 2006, when smart match was axiomatically symmetrical:
32 L<http://groups.google.com/group/perl.perl6.language/msg/bf2b486f089ad021>
34 Since then the Perl 6 target moved, but the Perl 5 implementation did not.
36 So it would be useful for someone to compare the Perl 6 smartmatch table
37 as of February 2006 L<http://svn.perl.org/viewvc/perl6/doc/trunk/design/syn/S03.pod?view=markup&pathrev=7615>
38 and the current table L<http://svn.perl.org/viewvc/perl6/doc/trunk/design/syn/S03.pod?revision=14556&view=markup>
39 and tabulate the differences in Perl 6. The annotated view of changes is
40 L<http://svn.perl.org/viewvc/perl6/doc/trunk/design/syn/S03.pod?view=annotate> and the diff is
41 C<svn diff -r7615:14556 http://svn.perl.org/perl6/doc/trunk/design/syn/S03.pod>
42 -- search for C<=head1 Smart matching>. (In theory F<viewvc> can generate that,
43 but in practice when I tried it hung forever, I assume "thinking")
45 With that done and published, someone (else) can then map any changed Perl 6
46 semantics back to Perl 5, based on how the existing semantics map to Perl 5:
47 L<http://search.cpan.org/~rgarcia/perl-5.10.0/pod/perlsyn.pod#Smart_matching_in_detail>
50 There are also some questions that need answering:
56 How do you negate one? (documentation issue)
57 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-01/msg00071.html
62 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-12/msg00799.html
64 * Should smart matches be symmetrical? (Perl 6 says no)
66 * Other differences between Perl 5 and Perl 6 smart match?
70 Objects and smart match
71 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-12/msg00865.html
75 =head2 Remove duplication of test setup.
77 Schwern notes, that there's duplication of code - lots and lots of tests have
78 some variation on the big block of C<$Is_Foo> checks. We can safely put this
79 into a file, change it to build an C<%Is> hash and require it. Maybe just put
80 it into F<test.pl>. Throw in the handy tainting subroutines.
82 =head2 POD -E<gt> HTML conversion in the core still sucks
84 Which is crazy given just how simple POD purports to be, and how simple HTML
85 can be. It's not actually I<as> simple as it sounds, particularly with the
86 flexibility POD allows for C<=item>, but it would be good to improve the
87 visual appeal of the HTML generated, and to avoid it having any validation
88 errors. See also L</make HTML install work>, as the layout of installation tree
89 is needed to improve the cross-linking.
91 The addition of C<Pod::Simple> and its related modules may make this task
94 =head2 Parallel testing
96 (This probably impacts much more than the core: also the Test::Harness
97 and TAP::* modules on CPAN.)
99 All of the tests in F<t/> can now be run in parallel, if C<$ENV{TEST_JOBS}>
100 is set. However, tests within each directory in F<ext> and F<lib> are still
101 run in series, with directories run in parallel. This is an adequate
102 heuristic, but it might be possible to relax it further, and get more
103 throughput. Specifically, it would be good to audit all of F<lib/*.t>, and
104 make them use C<File::Temp>.
106 =head2 Make Schwern poorer
108 We should have tests for everything. When all the core's modules are tested,
109 Schwern has promised to donate to $500 to TPF. We may need volunteers to
110 hold him upside down and shake vigorously in order to actually extract the
113 =head2 Improve the coverage of the core tests
115 Use Devel::Cover to ascertain the core modules's test coverage, then add
116 tests that are currently missing.
120 A full test suite for the B module would be nice.
122 =head2 A decent benchmark
124 C<perlbench> seems impervious to any recent changes made to the perl core. It
125 would be useful to have a reasonable general benchmarking suite that roughly
126 represented what current perl programs do, and measurably reported whether
127 tweaks to the core improve, degrade or don't really affect performance, to
128 guide people attempting to optimise the guts of perl. Gisle would welcome
129 new tests for perlbench.
131 =head2 fix tainting bugs
133 Fix the bugs revealed by running the test suite with the C<-t> switch (via
134 C<make test.taintwarn>).
136 =head2 Dual life everything
138 As part of the "dists" plan, anything that doesn't belong in the smallest perl
139 distribution needs to be dual lifed. Anything else can be too. Figure out what
140 changes would be needed to package that module and its tests up for CPAN, and
141 do so. Test it with older perl releases, and fix the problems you find.
143 To make a minimal perl distribution, it's useful to look at
144 F<t/lib/commonsense.t>.
146 =head2 Bundle dual life modules in ext/
148 For maintenance (and branch merging) reasons, it would be useful to move
149 some architecture-independent dual-life modules from lib/ to ext/, if this
150 has no negative impact on the build of perl itself.
152 =head2 POSIX memory footprint
154 Ilya observed that use POSIX; eats memory like there's no tomorrow, and at
155 various times worked to cut it down. There is probably still fat to cut out -
156 for example POSIX passes Exporter some very memory hungry data structures.
158 =head2 embed.pl/makedef.pl
160 There is a script F<embed.pl> that generates several header files to prefix
161 all of Perl's symbols in a consistent way, to provide some semblance of
162 namespace support in C<C>. Functions are declared in F<embed.fnc>, variables
163 in F<interpvar.h>. Quite a few of the functions and variables
164 are conditionally declared there, using C<#ifdef>. However, F<embed.pl>
165 doesn't understand the C macros, so the rules about which symbols are present
166 when is duplicated in F<makedef.pl>. Writing things twice is bad, m'kay.
167 It would be good to teach C<embed.pl> to understand the conditional
168 compilation, and hence remove the duplication, and the mistakes it has caused.
170 =head2 use strict; and AutoLoad
172 Currently if you write
175 use AutoLoader 'AUTOLOAD';
180 print join (' ', No, strict, here), "!\n";
183 then C<use strict;> isn't in force within the autoloaded subroutines. It would
184 be more consistent (and less surprising) to arrange for all lexical pragmas
185 in force at the __END__ block to be in force within each autoloaded subroutine.
187 There's a similar problem with SelfLoader.
189 =head2 profile installman
191 The F<installman> script is slow. All it is doing text processing, which we're
192 told is something Perl is good at. So it would be nice to know what it is doing
193 that is taking so much CPU, and where possible address it.
196 =head1 Tasks that need a little sysadmin-type knowledge
198 Or if you prefer, tasks that you would learn from, and broaden your skills
201 =head2 make HTML install work
203 There is an C<installhtml> target in the Makefile. It's marked as
204 "experimental". It would be good to get this tested, make it work reliably, and
205 remove the "experimental" tag. This would include
211 Checking that cross linking between various parts of the documentation works.
212 In particular that links work between the modules (files with POD in F<lib/>)
213 and the core documentation (files in F<pod/>)
217 Work out how to split C<perlfunc> into chunks, preferably one per function
218 group, preferably with general case code that could be used elsewhere.
219 Challenges here are correctly identifying the groups of functions that go
220 together, and making the right named external cross-links point to the right
221 page. Things to be aware of are C<-X>, groups such as C<getpwnam> to
222 C<endservent>, two or more C<=items> giving the different parameter lists, such
225 =item substr EXPR,OFFSET,LENGTH,REPLACEMENT
226 =item substr EXPR,OFFSET,LENGTH
227 =item substr EXPR,OFFSET
229 and different parameter lists having different meanings. (eg C<select>)
233 =head2 compressed man pages
235 Be able to install them. This would probably need a configure test to see how
236 the system does compressed man pages (same directory/different directory?
237 same filename/different filename), as well as tweaking the F<installman> script
238 to compress as necessary.
240 =head2 Add a code coverage target to the Makefile
242 Make it easy for anyone to run Devel::Cover on the core's tests. The steps
243 to do this manually are roughly
249 do a normal C<Configure>, but include Devel::Cover as a module to install
250 (see F<INSTALL> for how to do this)
258 cd t; HARNESS_PERL_SWITCHES=-MDevel::Cover ./perl -I../lib harness
262 Process the resulting Devel::Cover database
266 This just give you the coverage of the F<.pm>s. To also get the C level
273 Additionally tell C<Configure> to use the appropriate C compiler flags for
280 (instead of C<make perl>)
284 After running the tests run C<gcov> to generate all the F<.gcov> files.
285 (Including down in the subdirectories of F<ext/>
289 (From the top level perl directory) run C<gcov2perl> on all the C<.gcov> files
290 to get their stats into the cover_db directory.
294 Then process the Devel::Cover database
298 It would be good to add a single switch to C<Configure> to specify that you
299 wanted to perform perl level coverage, and another to specify C level
300 coverage, and have C<Configure> and the F<Makefile> do all the right things
303 =head2 Make Config.pm cope with differences between built and installed perl
305 Quite often vendors ship a perl binary compiled with their (pay-for)
306 compilers. People install a free compiler, such as gcc. To work out how to
307 build extensions, Perl interrogates C<%Config>, so in this situation
308 C<%Config> describes compilers that aren't there, and extension building
309 fails. This forces people into choosing between re-compiling perl themselves
310 using the compiler they have, or only using modules that the vendor ships.
312 It would be good to find a way teach C<Config.pm> about the installation setup,
313 possibly involving probing at install time or later, so that the C<%Config> in
314 a binary distribution better describes the installed machine, when the
315 installed machine differs from the build machine in some significant way.
317 =head2 linker specification files
319 Some platforms mandate that you provide a list of a shared library's external
320 symbols to the linker, so the core already has the infrastructure in place to
321 do this for generating shared perl libraries. My understanding is that the
322 GNU toolchain can accept an optional linker specification file, and restrict
323 visibility just to symbols declared in that file. It would be good to extend
324 F<makedef.pl> to support this format, and to provide a means within
325 C<Configure> to enable it. This would allow Unix users to test that the
326 export list is correct, and to build a perl that does not pollute the global
327 namespace with private symbols.
329 =head2 Cross-compile support
331 Currently C<Configure> understands C<-Dusecrosscompile> option. This option
332 arranges for building C<miniperl> for TARGET machine, so this C<miniperl> is
333 assumed then to be copied to TARGET machine and used as a replacement of full
336 This could be done little differently. Namely C<miniperl> should be built for
337 HOST and then full C<perl> with extensions should be compiled for TARGET.
338 This, however, might require extra trickery for %Config: we have one config
339 first for HOST and then another for TARGET. Tools like MakeMaker will be
340 mightily confused. Having around two different types of executables and
341 libraries (HOST and TARGET) makes life interesting for Makefiles and
342 shell (and Perl) scripts. There is $Config{run}, normally empty, which
343 can be used as an execution wrapper. Also note that in some
344 cross-compilation/execution environments the HOST and the TARGET do
345 not see the same filesystem(s), the $Config{run} may need to do some
346 file/directory copying back and forth.
350 Make F<pod/roffitall> be updated by F<pod/buildtoc>.
352 =head2 Split "linker" from "compiler"
354 Right now, Configure probes for two commands, and sets two variables:
358 =item * C<cc> (in F<cc.U>)
360 This variable holds the name of a command to execute a C compiler which
361 can resolve multiple global references that happen to have the same
362 name. Usual values are F<cc> and F<gcc>.
363 Fervent ANSI compilers may be called F<c89>. AIX has F<xlc>.
365 =item * C<ld> (in F<dlsrc.U>)
367 This variable indicates the program to be used to link
368 libraries for dynamic loading. On some systems, it is F<ld>.
369 On ELF systems, it should be C<$cc>. Mostly, we'll try to respect
370 the hint file setting.
374 There is an implicit historical assumption from around Perl5.000alpha
375 something, that C<$cc> is also the correct command for linking object files
376 together to make an executable. This may be true on Unix, but it's not true
377 on other platforms, and there are a maze of work arounds in other places (such
378 as F<Makefile.SH>) to cope with this.
380 Ideally, we should create a new variable to hold the name of the executable
381 linker program, probe for it in F<Configure>, and centralise all the special
382 case logic there or in hints files.
384 A small bikeshed issue remains - what to call it, given that C<$ld> is already
385 taken (arguably for the wrong thing now, but on SunOS 4.1 it is the command
386 for creating dynamically-loadable modules) and C<$link> could be confused with
387 the Unix command line executable of the same name, which does something
388 completely different. Andy Dougherty makes the counter argument "In parrot, I
389 tried to call the command used to link object files and libraries into an
390 executable F<link>, since that's what my vaguely-remembered DOS and VMS
391 experience suggested. I don't think any real confusion has ensued, so it's
392 probably a reasonable name for perl5 to use."
394 "Alas, I've always worried that introducing it would make things worse,
395 since now the module building utilities would have to look for
396 C<$Config{link}> and institute a fall-back plan if it weren't found."
397 Although I can see that as confusing, given that C<$Config{d_link}> is true
398 when (hard) links are available.
400 =head1 Tasks that need a little C knowledge
402 These tasks would need a little C knowledge, but don't need any specific
403 background or experience with XS, or how the Perl interpreter works
405 =head2 Weed out needless PERL_UNUSED_ARG
407 The C code uses the macro C<PERL_UNUSED_ARG> to stop compilers warning about
408 unused arguments. Often the arguments can't be removed, as there is an
409 external constraint that determines the prototype of the function, so this
410 approach is valid. However, there are some cases where C<PERL_UNUSED_ARG>
411 could be removed. Specifically
417 The prototypes of (nearly all) static functions can be changed
421 Unused arguments generated by short cut macros are wasteful - the short cut
422 macro used can be changed.
426 =head2 Modernize the order of directories in @INC
428 The way @INC is laid out by default, one cannot upgrade core (dual-life)
429 modules without overwriting files. This causes problems for binary
430 package builders. One possible proposal is laid out in this
432 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2002-04/msg02380.html>.
436 Natively 64-bit systems need neither -Duse64bitint nor -Duse64bitall.
437 On these systems, it might be the default compilation mode, and there
438 is currently no guarantee that passing no use64bitall option to the
439 Configure process will build a 32bit perl. Implementing -Duse32bit*
440 options would be nice for perl 5.12.
442 =head2 Profile Perl - am I hot or not?
444 The Perl source code is stable enough that it makes sense to profile it,
445 identify and optimise the hotspots. It would be good to measure the
446 performance of the Perl interpreter using free tools such as cachegrind,
447 gprof, and dtrace, and work to reduce the bottlenecks they reveal.
449 As part of this, the idea of F<pp_hot.c> is that it contains the I<hot> ops,
450 the ops that are most commonly used. The idea is that by grouping them, their
451 object code will be adjacent in the executable, so they have a greater chance
452 of already being in the CPU cache (or swapped in) due to being near another op
455 Except that it's not clear if these really are the most commonly used ops. So
456 as part of exercising your skills with coverage and profiling tools you might
457 want to determine what ops I<really> are the most commonly used. And in turn
458 suggest evictions and promotions to achieve a better F<pp_hot.c>.
460 One piece of Perl code that might make a good testbed is F<installman>.
462 =head2 Allocate OPs from arenas
464 Currently all new OP structures are individually malloc()ed and free()d.
465 All C<malloc> implementations have space overheads, and are now as fast as
466 custom allocates so it would both use less memory and less CPU to allocate
467 the various OP structures from arenas. The SV arena code can probably be
470 Note that Configuring perl with C<-Accflags=-DPL_OP_SLAB_ALLOC> will use
471 Perl_Slab_alloc() to pack optrees into a contiguous block, which is
472 probably superior to the use of OP arenas, esp. from a cache locality
473 standpoint. See L<Profile Perl - am I hot or not?>.
475 =head2 Improve win32/wince.c
477 Currently, numerous functions look virtually, if not completely,
478 identical in both C<win32/wince.c> and C<win32/win32.c> files, which can't
481 =head2 Use secure CRT functions when building with VC8 on Win32
483 Visual C++ 2005 (VC++ 8.x) deprecated a number of CRT functions on the basis
484 that they were "unsafe" and introduced differently named secure versions of
485 them as replacements, e.g. instead of writing
487 FILE* f = fopen(__FILE__, "r");
492 errno_t err = fopen_s(&f, __FILE__, "r");
494 Currently, the warnings about these deprecations have been disabled by adding
495 -D_CRT_SECURE_NO_DEPRECATE to the CFLAGS. It would be nice to remove that
496 warning suppressant and actually make use of the new secure CRT functions.
498 There is also a similar issue with POSIX CRT function names like fileno having
499 been deprecated in favour of ISO C++ conformant names like _fileno. These
500 warnings are also currently suppressed by adding -D_CRT_NONSTDC_NO_DEPRECATE. It
501 might be nice to do as Microsoft suggest here too, although, unlike the secure
502 functions issue, there is presumably little or no benefit in this case.
504 =head2 Fix POSIX::access() and chdir() on Win32
506 These functions currently take no account of DACLs and therefore do not behave
507 correctly in situations where access is restricted by DACLs (as opposed to the
508 read-only attribute).
510 Furthermore, POSIX::access() behaves differently for directories having the
511 read-only attribute set depending on what CRT library is being used. For
512 example, the _access() function in the VC6 and VC7 CRTs (wrongly) claim that
513 such directories are not writable, whereas in fact all directories are writable
514 unless access is denied by DACLs. (In the case of directories, the read-only
515 attribute actually only means that the directory cannot be deleted.) This CRT
516 bug is fixed in the VC8 and VC9 CRTs (but, of course, the directory may still
517 not actually be writable if access is indeed denied by DACLs).
519 For the chdir() issue, see ActiveState bug #74552:
520 http://bugs.activestate.com/show_bug.cgi?id=74552
522 Therefore, DACLs should be checked both for consistency across CRTs and for
525 (Note that perl's -w operator should not be modified to check DACLs. It has
526 been written so that it reflects the state of the read-only attribute, even
527 for directories (whatever CRT is being used), for symmetry with chmod().)
529 =head2 strcat(), strcpy(), strncat(), strncpy(), sprintf(), vsprintf()
531 Maybe create a utility that checks after each libperl.a creation that
532 none of the above (nor sprintf(), vsprintf(), or *SHUDDER* gets())
533 ever creep back to libperl.a.
535 nm libperl.a | ./miniperl -alne '$o = $F[0] if /:$/; print "$o $F[1]" if $F[0] eq "U" && $F[1] =~ /^(?:strn?c(?:at|py)|v?sprintf|gets)$/'
537 Note, of course, that this will only tell whether B<your> platform
538 is using those naughty interfaces.
540 =head2 -D_FORTIFY_SOURCE=2, -fstack-protector
542 Recent glibcs support C<-D_FORTIFY_SOURCE=2> and recent gcc
543 (4.1 onwards?) supports C<-fstack-protector>, both of which give
544 protection against various kinds of buffer overflow problems.
545 These should probably be used for compiling Perl whenever available,
546 Configure and/or hints files should be adjusted to probe for the
547 availability of these features and enable them as appropriate.
549 =head2 Arenas for GPs? For MAGIC?
551 C<struct gp> and C<struct magic> are both currently allocated by C<malloc>.
552 It might be a speed or memory saving to change to using arenas. Or it might
553 not. It would need some suitable benchmarking first. In particular, C<GP>s
554 can probably be changed with minimal compatibility impact (probably nothing
555 outside of the core, or even outside of F<gv.c> allocates them), but they
556 probably aren't allocated/deallocated often enough for a speed saving. Whereas
557 C<MAGIC> is allocated/deallocated more often, but in turn, is also something
558 more externally visible, so changing the rules here may bite external code.
562 Several SV body structs are now the same size, notably PVMG and PVGV, PVAV and
563 PVHV, and PVCV and PVFM. It should be possible to allocate and return same
564 sized bodies from the same actual arena, rather than maintaining one arena for
565 each. This could save 4-6K per thread, of memory no longer tied up in the
566 not-yet-allocated part of an arena.
569 =head1 Tasks that need a knowledge of XS
571 These tasks would need C knowledge, and roughly the level of knowledge of
572 the perl API that comes from writing modules that use XS to interface to
575 =head2 safely supporting POSIX SA_SIGINFO
577 Some years ago Jarkko supplied patches to provide support for the POSIX
578 SA_SIGINFO feature in Perl, passing the extra data to the Perl signal handler.
580 Unfortunately, it only works with "unsafe" signals, because under safe
581 signals, by the time Perl gets to run the signal handler, the extra
582 information has been lost. Moreover, it's not easy to store it somewhere,
583 as you can't call mutexs, or do anything else fancy, from inside a signal
586 So it strikes me that we could provide safe SA_SIGINFO support
592 Provide global variables for two file descriptors
596 When the first request is made via C<sigaction> for C<SA_SIGINFO>, create a
597 pipe, store the reader in one, the writer in the other
601 In the "safe" signal handler (C<Perl_csighandler()>/C<S_raise_signal()>), if
602 the C<siginfo_t> pointer non-C<NULL>, and the writer file handle is open,
608 serialise signal number, C<struct siginfo_t> (or at least the parts we care
609 about) into a small auto char buff
613 C<write()> that (non-blocking) to the writer fd
619 if it writes 100%, flag the signal in a counter of "signals on the pipe" akin
620 to the current per-signal-number counts
624 if it writes 0%, assume the pipe is full. Flag the data as lost?
628 if it writes partially, croak a panic, as your OS is broken.
636 in the regular C<PERL_ASYNC_CHECK()> processing, if there are "signals on
637 the pipe", read the data out, deserialise, build the Perl structures on
638 the stack (code in C<Perl_sighandler()>, the "unsafe" handler), and call as
643 I think that this gets us decent C<SA_SIGINFO> support, without the current risk
644 of running Perl code inside the signal handler context. (With all the dangers
645 of things like C<malloc> corruption that that currently offers us)
647 For more information see the thread starting with this message:
648 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-03/msg00305.html
650 =head2 autovivification
652 Make all autovivification consistent w.r.t LVALUE/RVALUE and strict/no strict;
654 This task is incremental - even a little bit of work on it will help.
656 =head2 Unicode in Filenames
658 chdir, chmod, chown, chroot, exec, glob, link, lstat, mkdir, open,
659 opendir, qx, readdir, readlink, rename, rmdir, stat, symlink, sysopen,
660 system, truncate, unlink, utime, -X. All these could potentially accept
661 Unicode filenames either as input or output (and in the case of system
662 and qx Unicode in general, as input or output to/from the shell).
663 Whether a filesystem - an operating system pair understands Unicode in
666 Known combinations that have some level of understanding include
667 Microsoft NTFS, Apple HFS+ (In Mac OS 9 and X) and Apple UFS (in Mac
668 OS X), NFS v4 is rumored to be Unicode, and of course Plan 9. How to
669 create Unicode filenames, what forms of Unicode are accepted and used
670 (UCS-2, UTF-16, UTF-8), what (if any) is the normalization form used,
671 and so on, varies. Finding the right level of interfacing to Perl
672 requires some thought. Remember that an OS does not implicate a
675 (The Windows -C command flag "wide API support" has been at least
676 temporarily retired in 5.8.1, and the -C has been repurposed, see
679 Most probably the right way to do this would be this:
680 L</"Virtualize operating system access">.
682 =head2 Unicode in %ENV
684 Currently the %ENV entries are always byte strings.
685 See L</"Virtualize operating system access">.
687 =head2 Unicode and glob()
689 Currently glob patterns and filenames returned from File::Glob::glob()
690 are always byte strings. See L</"Virtualize operating system access">.
692 =head2 Unicode and lc/uc operators
694 Some built-in operators (C<lc>, C<uc>, etc.) behave differently, based on
695 what the internal encoding of their argument is. That should not be the
696 case. Maybe add a pragma to switch behaviour.
698 =head2 use less 'memory'
700 Investigate trade offs to switch out perl's choices on memory usage.
701 Particularly perl should be able to give memory back.
703 This task is incremental - even a little bit of work on it will help.
705 =head2 Re-implement C<:unique> in a way that is actually thread-safe
707 The old implementation made bad assumptions on several levels. A good 90%
708 solution might be just to make C<:unique> work to share the string buffer
709 of SvPVs. That way large constant strings can be shared between ithreads,
710 such as the configuration information in F<Config>.
712 =head2 Make tainting consistent
714 Tainting would be easier to use if it didn't take documented shortcuts and
715 allow taint to "leak" everywhere within an expression.
717 =head2 readpipe(LIST)
719 system() accepts a LIST syntax (and a PROGRAM LIST syntax) to avoid
720 running a shell. readpipe() (the function behind qx//) could be similarly
723 =head2 Audit the code for destruction ordering assumptions
727 /* Need to check SvMAGICAL, as during global destruction it may be that
728 AvARYLEN(av) has been freed before av, and hence the SvANY() pointer
729 is now part of the linked list of SV heads, rather than pointing to
730 the original body. */
731 /* FIXME - audit the code for other bugs like this one. */
733 adding the C<SvMAGICAL> check to
735 if (AvARYLEN(av) && SvMAGICAL(AvARYLEN(av))) {
736 MAGIC *mg = mg_find (AvARYLEN(av), PERL_MAGIC_arylen);
738 Go through the core and look for similar assumptions that SVs have particular
739 types, as all bets are off during global destruction.
741 =head2 Extend PerlIO and PerlIO::Scalar
743 PerlIO::Scalar doesn't know how to truncate(). Implementing this
744 would require extending the PerlIO vtable.
746 Similarly the PerlIO vtable doesn't know about formats (write()), or
747 about stat(), or chmod()/chown(), utime(), or flock().
749 (For PerlIO::Scalar it's hard to see what e.g. mode bits or ownership
752 PerlIO doesn't do directories or symlinks, either: mkdir(), rmdir(),
753 opendir(), closedir(), seekdir(), rewinddir(), glob(); symlink(),
756 See also L</"Virtualize operating system access">.
758 =head2 -C on the #! line
760 It should be possible to make -C work correctly if found on the #! line,
761 given that all perl command line options are strict ASCII, and -C changes
762 only the interpretation of non-ASCII characters, and not for the script file
763 handle. To make it work needs some investigation of the ordering of function
764 calls during startup, and (by implication) a bit of tweaking of that order.
766 =head2 Organize error messages
768 Perl's diagnostics (error messages, see L<perldiag>) could use
769 reorganizing and formalizing so that each error message has its
770 stable-for-all-eternity unique id, categorized by severity, type, and
771 subsystem. (The error messages would be listed in a datafile outside
772 of the Perl source code, and the source code would only refer to the
773 messages by the id.) This clean-up and regularizing should apply
774 for all croak() messages.
776 This would enable all sorts of things: easier translation/localization
777 of the messages (though please do keep in mind the caveats of
778 L<Locale::Maketext> about too straightforward approaches to
779 translation), filtering by severity, and instead of grepping for a
780 particular error message one could look for a stable error id. (Of
781 course, changing the error messages by default would break all the
782 existing software depending on some particular error message...)
784 This kind of functionality is known as I<message catalogs>. Look for
785 inspiration for example in the catgets() system, possibly even use it
786 if available-- but B<only> if available, all platforms will B<not>
789 For the really pure at heart, consider extending this item to cover
790 also the warning messages (see L<perllexwarn>, C<warnings.pl>).
792 =head1 Tasks that need a knowledge of the interpreter
794 These tasks would need C knowledge, and knowledge of how the interpreter works,
795 or a willingness to learn.
797 =head2 error reporting of [$a ; $b]
799 Using C<;> inside brackets is a syntax error, and we don't propose to change
800 that by giving it any meaning. However, it's not reported very helpfully:
802 $ perl -e '$a = [$b; $c];'
803 syntax error at -e line 1, near "$b;"
804 syntax error at -e line 1, near "$c]"
805 Execution of -e aborted due to compilation errors.
807 It should be possible to hook into the tokeniser or the lexer, so that when a
808 C<;> is parsed where it is not legal as a statement terminator (ie inside
809 C<{}> used as a hashref, C<[]> or C<()>) it issues an error something like
810 I<';' isn't legal inside an expression - if you need multiple statements use a
811 do {...} block>. See the thread starting at
812 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-09/msg00573.html
814 =head2 lexicals used only once
818 $ perl -we '$pie = 42'
819 Name "main::pie" used only once: possible typo at -e line 1.
823 $ perl -we 'my $pie = 42'
825 Logically all lexicals used only once should warn, if the user asks for
826 warnings. An unworked RT ticket (#5087) has been open for almost seven
827 years for this discrepancy.
831 The handling of Unicode is unclean in many places. For example, the regexp
832 engine matches in Unicode semantics whenever the string or the pattern is
833 flagged as UTF-8, but that should not be dependent on an internal storage
834 detail of the string. Likewise, case folding behaviour is dependent on the
835 UTF8 internal flag being on or off.
837 =head2 Properly Unicode safe tokeniser and pads.
839 The tokeniser isn't actually very UTF-8 clean. C<use utf8;> is a hack -
840 variable names are stored in stashes as raw bytes, without the utf-8 flag
841 set. The pad API only takes a C<char *> pointer, so that's all bytes too. The
842 tokeniser ignores the UTF-8-ness of C<PL_rsfp>, or any SVs returned from
843 source filters. All this could be fixed.
845 =head2 state variable initialization in list context
847 Currently this is illegal:
849 state ($a, $b) = foo();
851 In Perl 6, C<state ($a) = foo();> and C<(state $a) = foo();> have different
852 semantics, which is tricky to implement in Perl 5 as currently they produce
853 the same opcode trees. The Perl 6 design is firm, so it would be good to
854 implement the necessary code in Perl 5. There are comments in
855 C<Perl_newASSIGNOP()> that show the code paths taken by various assignment
856 constructions involving state variables.
858 =head2 Implement $value ~~ 0 .. $range
860 It would be nice to extend the syntax of the C<~~> operator to also
861 understand numeric (and maybe alphanumeric) ranges.
863 =head2 A does() built-in
865 Like ref(), only useful. It would call the C<DOES> method on objects; it
866 would also tell whether something can be dereferenced as an
867 array/hash/etc., or used as a regexp, etc.
868 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-03/msg00481.html>
870 =head2 Tied filehandles and write() don't mix
872 There is no method on tied filehandles to allow them to be called back by
875 =head2 Attach/detach debugger from running program
877 The old perltodo notes "With C<gdb>, you can attach the debugger to a running
878 program if you pass the process ID. It would be good to do this with the Perl
879 debugger on a running Perl program, although I'm not sure how it would be
880 done." ssh and screen do this with named pipes in /tmp. Maybe we can too.
882 =head2 Optimize away empty destructors
884 Defining an empty DESTROY method might be useful (notably in
885 AUTOLOAD-enabled classes), but it's still a bit expensive to call. That
886 could probably be optimized.
888 =head2 LVALUE functions for lists
890 The old perltodo notes that lvalue functions don't work for list or hash
891 slices. This would be good to fix.
893 =head2 regexp optimiser optional
895 The regexp optimiser is not optional. It should configurable to be, to allow
896 its performance to be measured, and its bugs to be easily demonstrated.
898 =head2 delete &function
900 Allow to delete functions. One can already undef them, but they're still
903 =head2 C</w> regex modifier
905 That flag would enable to match whole words, and also to interpolate
906 arrays as alternations. With it, C</P/w> would be roughly equivalent to:
908 do { local $"='|'; /\b(?:P)\b/ }
910 See L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-01/msg00400.html>
913 =head2 optional optimizer
915 Make the peephole optimizer optional. Currently it performs two tasks as
916 it walks the optree - genuine peephole optimisations, and necessary fixups of
917 ops. It would be good to find an efficient way to switch out the
918 optimisations whilst keeping the fixups.
920 =head2 You WANT *how* many
922 Currently contexts are void, scalar and list. split has a special mechanism in
923 place to pass in the number of return values wanted. It would be useful to
924 have a general mechanism for this, backwards compatible and little speed hit.
925 This would allow proposals such as short circuiting sort to be implemented
928 =head2 lexical aliases
930 Allow lexical aliases (maybe via the syntax C<my \$alias = \$foo>.
932 =head2 entersub XS vs Perl
934 At the moment pp_entersub is huge, and has code to deal with entering both
935 perl and XS subroutines. Subroutine implementations rarely change between
936 perl and XS at run time, so investigate using 2 ops to enter subs (one for
937 XS, one for perl) and swap between if a sub is redefined.
941 Self-ties are currently illegal because they caused too many segfaults. Maybe
942 the causes of these could be tracked down and self-ties on all types
945 =head2 Optimize away @_
947 The old perltodo notes "Look at the "reification" code in C<av.c>".
949 =head2 Virtualize operating system access
951 Implement a set of "vtables" that virtualizes operating system access
952 (open(), mkdir(), unlink(), readdir(), getenv(), etc.) At the very
953 least these interfaces should take SVs as "name" arguments instead of
954 bare char pointers; probably the most flexible and extensible way
955 would be for the Perl-facing interfaces to accept HVs. The system
956 needs to be per-operating-system and per-file-system
957 hookable/filterable, preferably both from XS and Perl level
958 (L<perlport/"Files and Filesystems"> is good reading at this point,
959 in fact, all of L<perlport> is.)
961 This has actually already been implemented (but only for Win32),
962 take a look at F<iperlsys.h> and F<win32/perlhost.h>. While all Win32
963 variants go through a set of "vtables" for operating system access,
964 non-Win32 systems currently go straight for the POSIX/UNIX-style
965 system/library call. Similar system as for Win32 should be
966 implemented for all platforms. The existing Win32 implementation
967 probably does not need to survive alongside this proposed new
968 implementation, the approaches could be merged.
970 What would this give us? One often-asked-for feature this would
971 enable is using Unicode for filenames, and other "names" like %ENV,
972 usernames, hostnames, and so forth.
973 (See L<perlunicode/"When Unicode Does Not Happen">.)
975 But this kind of virtualization would also allow for things like
976 virtual filesystems, virtual networks, and "sandboxes" (though as long
977 as dynamic loading of random object code is allowed, not very safe
978 sandboxes since external code of course know not of Perl's vtables).
979 An example of a smaller "sandbox" is that this feature can be used to
980 implement per-thread working directories: Win32 already does this.
982 See also L</"Extend PerlIO and PerlIO::Scalar">.
984 =head2 Investigate PADTMP hash pessimisation
986 The peephole optimiser converts constants used for hash key lookups to shared
987 hash key scalars. Under ithreads, something is undoing this work.
988 See http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-09/msg00793.html
990 =head2 Store the current pad in the OP slab allocator
993 I hope that I got that "current pad" part correct
995 Currently we leak ops in various cases of parse failure. I suggested that we
996 could solve this by always using the op slab allocator, and walking it to
997 free ops. Dave comments that as some ops are already freed during optree
998 creation one would have to mark which ops are freed, and not double free them
999 when walking the slab. He notes that one problem with this is that for some ops
1000 you have to know which pad was current at the time of allocation, which does
1001 change. I suggested storing a pointer to the current pad in the memory allocated
1002 for the slab, and swapping to a new slab each time the pad changes. Dave thinks
1003 that this would work.
1005 =head2 repack the optree
1007 Repacking the optree after execution order is determined could allow
1008 removal of NULL ops, and optimal ordering of OPs with respect to cache-line
1009 filling. The slab allocator could be reused for this purpose. I think that
1010 the best way to do this is to make it an optional step just before the
1011 completed optree is attached to anything else, and to use the slab allocator
1012 unchanged, so that freeing ops is identical whether or not this step runs.
1013 Note that the slab allocator allocates ops downwards in memory, so one would
1014 have to actually "allocate" the ops in reverse-execution order to get them
1015 contiguous in memory in execution order.
1017 See http://www.nntp.perl.org/group/perl.perl5.porters/2007/12/msg131975.html
1019 Note that running this copy, and then freeing all the old location ops would
1020 cause their slabs to be freed, which would eliminate possible memory wastage if
1021 the previous suggestion is implemented, and we swap slabs more frequently.
1023 =head2 eliminate incorrect line numbers in warnings
1031 } elsif ($undef == 0) {
1034 used to produce this output:
1036 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1037 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1039 where the line of the second warning was misreported - it should be line 5.
1040 Rafael fixed this - the problem arose because there was no nextstate OP
1041 between the execution of the C<if> and the C<elsif>, hence C<PL_curcop> still
1042 reports that the currently executing line is line 4. The solution was to inject
1043 a nextstate OPs for each C<elsif>, although it turned out that the nextstate
1044 OP needed to be a nulled OP, rather than a live nextstate OP, else other line
1045 numbers became misreported. (Jenga!)
1047 The problem is more general than C<elsif> (although the C<elsif> case is the
1048 most common and the most confusing). Ideally this code
1058 would produce this output
1060 Use of uninitialized value $undef in addition (+) at wrong.pl line 4.
1061 Use of uninitialized value $undef in addition (+) at wrong.pl line 7.
1063 (rather than lines 4 and 5), but this would seem to require every OP to carry
1064 (at least) line number information.
1066 What might work is to have an optional line number in memory just before the
1067 BASEOP structure, with a flag bit in the op to say whether it's present.
1068 Initially during compile every OP would carry its line number. Then add a late
1069 pass to the optimiser (potentially combined with L</repack the optree>) which
1070 looks at the two ops on every edge of the graph of the execution path. If
1071 the line number changes, flags the destination OP with this information.
1072 Once all paths are traced, replace every op with the flag with a
1073 nextstate-light op (that just updates C<PL_curcop>), which in turn then passes
1074 control on to the true op. All ops would then be replaced by variants that
1075 do not store the line number. (Which, logically, why it would work best in
1076 conjunction with L</repack the optree>, as that is already copying/reallocating
1079 (Although I should note that we're not certain that doing this for the general
1082 =head2 optimize tail-calls
1084 Tail-calls present an opportunity for broadly applicable optimization;
1085 anywhere that C<< return foo(...) >> is called, the outer return can
1086 be replaced by a goto, and foo will return directly to the outer
1087 caller, saving (conservatively) 25% of perl's call&return cost, which
1088 is relatively higher than in C. The scheme language is known to do
1089 this heavily. B::Concise provides good insight into where this
1090 optimization is possible, ie anywhere entersub,leavesub op-sequence
1093 perl -MO=Concise,-exec,a,b,-main -e 'sub a{ 1 }; sub b {a()}; b(2)'
1095 Bottom line on this is probably a new pp_tailcall function which
1096 combines the code in pp_entersub, pp_leavesub. This should probably
1097 be done 1st in XS, and using B::Generate to patch the new OP into the
1102 Tasks that will get your name mentioned in the description of the "Highlights
1105 =head2 make ithreads more robust
1107 Generally make ithreads more robust. See also L</iCOW>
1109 This task is incremental - even a little bit of work on it will help, and
1110 will be greatly appreciated.
1112 One bit would be to write the missing code in sv.c:Perl_dirp_dup.
1114 Fix Perl_sv_dup, et al so that threads can return objects.
1118 Sarathy and Arthur have a proposal for an improved Copy On Write which
1119 specifically will be able to COW new ithreads. If this can be implemented
1120 it would be a good thing.
1122 =head2 (?{...}) closures in regexps
1124 Fix (or rewrite) the implementation of the C</(?{...})/> closures.
1126 =head2 A re-entrant regexp engine
1128 This will allow the use of a regex from inside (?{ }), (??{ }) and
1129 (?(?{ })|) constructs.
1131 =head2 Add class set operations to regexp engine
1133 Apparently these are quite useful. Anyway, Jeffery Friedl wants them.
1135 demerphq has this on his todo list, but right at the bottom.