3 perltodo - Perl TO-DO List
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:/pod/perltodo.pod>
10 The tasks we think are smaller or easier are listed first. Anyone is welcome
11 to work on any of these, but it's a good idea to first contact
12 I<perl5-porters@perl.org> to avoid duplication of effort, and to learn from
13 any previous attempts. By all means contact a pumpking privately first if you
16 Whilst patches to make the list shorter are most welcome, ideas to add to
17 the list are also encouraged. Check the perl5-porters archives for past
18 ideas, and any discussion about them. One set of archives may be found at
19 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/>
21 What can we offer you in return? Fame, fortune, and everlasting glory? Maybe
22 not, but if your patch is incorporated, then we'll add your name to the
23 F<AUTHORS> file, which ships in the official distribution. How many other
24 programming languages offer you 1 line of immortality?
26 =head1 Tasks that only need Perl knowledge
28 =head2 Migrate t/ from custom TAP generation
30 Many tests below F<t/> still generate TAP by "hand", rather than using library
31 functions. As explained in L<perlhack/TESTING>, tests in F<t/> are
32 written in a particular way to test that more complex constructions actually
33 work before using them routinely. Hence they don't use C<Test::More>, but
34 instead there is an intentionally simpler library, F<t/test.pl>. However,
35 quite a few tests in F<t/> have not been refactored to use it. Refactoring
36 any of these tests, one at a time, is a useful thing TODO.
38 The subdirectories F<base>, F<cmd> and F<comp>, that contain the most
39 basic tests, should be excluded from this task.
41 =head2 Automate perldelta generation
43 The perldelta file accompanying each release summaries the major changes.
44 It's mostly manually generated currently, but some of that could be
45 automated with a bit of perl, specifically the generation of
49 =item Modules and Pragmata
51 =item New Documentation
57 See F<Porting/how_to_write_a_perldelta.pod> for details.
59 =head2 Remove duplication of test setup.
61 Schwern notes, that there's duplication of code - lots and lots of tests have
62 some variation on the big block of C<$Is_Foo> checks. We can safely put this
63 into a file, change it to build an C<%Is> hash and require it. Maybe just put
64 it into F<test.pl>. Throw in the handy tainting subroutines.
66 =head2 POD -E<gt> HTML conversion in the core still sucks
68 Which is crazy given just how simple POD purports to be, and how simple HTML
69 can be. It's not actually I<as> simple as it sounds, particularly with the
70 flexibility POD allows for C<=item>, but it would be good to improve the
71 visual appeal of the HTML generated, and to avoid it having any validation
72 errors. See also L</make HTML install work>, as the layout of installation tree
73 is needed to improve the cross-linking.
75 The addition of C<Pod::Simple> and its related modules may make this task
78 =head2 Make Schwern poorer
80 We should have tests for everything. When all the core's modules are tested,
81 Schwern has promised to donate to $500 to TPF. We may need volunteers to
82 hold him upside down and shake vigorously in order to actually extract the
85 =head2 Improve the coverage of the core tests
87 Use Devel::Cover to ascertain the core modules' test coverage, then add
88 tests that are currently missing.
92 A full test suite for the B module would be nice.
94 =head2 A decent benchmark
96 C<perlbench> seems impervious to any recent changes made to the perl core. It
97 would be useful to have a reasonable general benchmarking suite that roughly
98 represented what current perl programs do, and measurably reported whether
99 tweaks to the core improve, degrade or don't really affect performance, to
100 guide people attempting to optimise the guts of perl. Gisle would welcome
101 new tests for perlbench.
103 =head2 fix tainting bugs
105 Fix the bugs revealed by running the test suite with the C<-t> switch (via
106 C<make test.taintwarn>).
108 =head2 Dual life everything
110 As part of the "dists" plan, anything that doesn't belong in the smallest perl
111 distribution needs to be dual lifed. Anything else can be too. Figure out what
112 changes would be needed to package that module and its tests up for CPAN, and
113 do so. Test it with older perl releases, and fix the problems you find.
115 To make a minimal perl distribution, it's useful to look at
116 F<t/lib/commonsense.t>.
118 =head2 POSIX memory footprint
120 Ilya observed that use POSIX; eats memory like there's no tomorrow, and at
121 various times worked to cut it down. There is probably still fat to cut out -
122 for example POSIX passes Exporter some very memory hungry data structures.
124 =head2 embed.pl/makedef.pl
126 There is a script F<embed.pl> that generates several header files to prefix
127 all of Perl's symbols in a consistent way, to provide some semblance of
128 namespace support in C<C>. Functions are declared in F<embed.fnc>, variables
129 in F<interpvar.h>. Quite a few of the functions and variables
130 are conditionally declared there, using C<#ifdef>. However, F<embed.pl>
131 doesn't understand the C macros, so the rules about which symbols are present
132 when is duplicated in F<makedef.pl>. Writing things twice is bad, m'kay.
133 It would be good to teach C<embed.pl> to understand the conditional
134 compilation, and hence remove the duplication, and the mistakes it has caused.
136 =head2 use strict; and AutoLoad
138 Currently if you write
141 use AutoLoader 'AUTOLOAD';
146 print join (' ', No, strict, here), "!\n";
149 then C<use strict;> isn't in force within the autoloaded subroutines. It would
150 be more consistent (and less surprising) to arrange for all lexical pragmas
151 in force at the __END__ block to be in force within each autoloaded subroutine.
153 There's a similar problem with SelfLoader.
155 =head2 profile installman
157 The F<installman> script is slow. All it is doing text processing, which we're
158 told is something Perl is good at. So it would be nice to know what it is doing
159 that is taking so much CPU, and where possible address it.
161 =head2 enable lexical enabling/disabling of individual warnings
163 Currently, warnings can only be enabled or disabled by category. There
164 are times when it would be useful to quash a single warning, not a
167 =head2 document diagnostics
169 Many diagnostic messages are not currently documented. The list is at the end
172 =head1 Tasks that need a little sysadmin-type knowledge
174 Or if you prefer, tasks that you would learn from, and broaden your skills
177 =head2 make HTML install work
179 There is an C<installhtml> target in the Makefile. It's marked as
180 "experimental". It would be good to get this tested, make it work reliably, and
181 remove the "experimental" tag. This would include
187 Checking that cross linking between various parts of the documentation works.
188 In particular that links work between the modules (files with POD in F<lib/>)
189 and the core documentation (files in F<pod/>)
193 Work out how to split C<perlfunc> into chunks, preferably one per function
194 group, preferably with general case code that could be used elsewhere.
195 Challenges here are correctly identifying the groups of functions that go
196 together, and making the right named external cross-links point to the right
197 page. Things to be aware of are C<-X>, groups such as C<getpwnam> to
198 C<endservent>, two or more C<=items> giving the different parameter lists, such
201 =item substr EXPR,OFFSET,LENGTH,REPLACEMENT
202 =item substr EXPR,OFFSET,LENGTH
203 =item substr EXPR,OFFSET
205 and different parameter lists having different meanings. (eg C<select>)
209 =head2 compressed man pages
211 Be able to install them. This would probably need a configure test to see how
212 the system does compressed man pages (same directory/different directory?
213 same filename/different filename), as well as tweaking the F<installman> script
214 to compress as necessary.
216 =head2 Add a code coverage target to the Makefile
218 Make it easy for anyone to run Devel::Cover on the core's tests. The steps
219 to do this manually are roughly
225 do a normal C<Configure>, but include Devel::Cover as a module to install
226 (see L<INSTALL> for how to do this)
234 cd t; HARNESS_PERL_SWITCHES=-MDevel::Cover ./perl -I../lib harness
238 Process the resulting Devel::Cover database
242 This just give you the coverage of the F<.pm>s. To also get the C level
249 Additionally tell C<Configure> to use the appropriate C compiler flags for
256 (instead of C<make perl>)
260 After running the tests run C<gcov> to generate all the F<.gcov> files.
261 (Including down in the subdirectories of F<ext/>
265 (From the top level perl directory) run C<gcov2perl> on all the C<.gcov> files
266 to get their stats into the cover_db directory.
270 Then process the Devel::Cover database
274 It would be good to add a single switch to C<Configure> to specify that you
275 wanted to perform perl level coverage, and another to specify C level
276 coverage, and have C<Configure> and the F<Makefile> do all the right things
279 =head2 Make Config.pm cope with differences between built and installed perl
281 Quite often vendors ship a perl binary compiled with their (pay-for)
282 compilers. People install a free compiler, such as gcc. To work out how to
283 build extensions, Perl interrogates C<%Config>, so in this situation
284 C<%Config> describes compilers that aren't there, and extension building
285 fails. This forces people into choosing between re-compiling perl themselves
286 using the compiler they have, or only using modules that the vendor ships.
288 It would be good to find a way teach C<Config.pm> about the installation setup,
289 possibly involving probing at install time or later, so that the C<%Config> in
290 a binary distribution better describes the installed machine, when the
291 installed machine differs from the build machine in some significant way.
293 =head2 linker specification files
295 Some platforms mandate that you provide a list of a shared library's external
296 symbols to the linker, so the core already has the infrastructure in place to
297 do this for generating shared perl libraries. My understanding is that the
298 GNU toolchain can accept an optional linker specification file, and restrict
299 visibility just to symbols declared in that file. It would be good to extend
300 F<makedef.pl> to support this format, and to provide a means within
301 C<Configure> to enable it. This would allow Unix users to test that the
302 export list is correct, and to build a perl that does not pollute the global
303 namespace with private symbols, and will fail in the same way as msvc or mingw
304 builds or when using PERL_DL_NONLAZY=1.
306 =head2 Cross-compile support
308 Currently C<Configure> understands C<-Dusecrosscompile> option. This option
309 arranges for building C<miniperl> for TARGET machine, so this C<miniperl> is
310 assumed then to be copied to TARGET machine and used as a replacement of full
313 This could be done little differently. Namely C<miniperl> should be built for
314 HOST and then full C<perl> with extensions should be compiled for TARGET.
315 This, however, might require extra trickery for %Config: we have one config
316 first for HOST and then another for TARGET. Tools like MakeMaker will be
317 mightily confused. Having around two different types of executables and
318 libraries (HOST and TARGET) makes life interesting for Makefiles and
319 shell (and Perl) scripts. There is $Config{run}, normally empty, which
320 can be used as an execution wrapper. Also note that in some
321 cross-compilation/execution environments the HOST and the TARGET do
322 not see the same filesystem(s), the $Config{run} may need to do some
323 file/directory copying back and forth.
327 Make F<pod/roffitall> be updated by F<pod/buildtoc>.
329 =head2 Split "linker" from "compiler"
331 Right now, Configure probes for two commands, and sets two variables:
335 =item * C<cc> (in F<cc.U>)
337 This variable holds the name of a command to execute a C compiler which
338 can resolve multiple global references that happen to have the same
339 name. Usual values are F<cc> and F<gcc>.
340 Fervent ANSI compilers may be called F<c89>. AIX has F<xlc>.
342 =item * C<ld> (in F<dlsrc.U>)
344 This variable indicates the program to be used to link
345 libraries for dynamic loading. On some systems, it is F<ld>.
346 On ELF systems, it should be C<$cc>. Mostly, we'll try to respect
347 the hint file setting.
351 There is an implicit historical assumption from around Perl5.000alpha
352 something, that C<$cc> is also the correct command for linking object files
353 together to make an executable. This may be true on Unix, but it's not true
354 on other platforms, and there are a maze of work arounds in other places (such
355 as F<Makefile.SH>) to cope with this.
357 Ideally, we should create a new variable to hold the name of the executable
358 linker program, probe for it in F<Configure>, and centralise all the special
359 case logic there or in hints files.
361 A small bikeshed issue remains - what to call it, given that C<$ld> is already
362 taken (arguably for the wrong thing now, but on SunOS 4.1 it is the command
363 for creating dynamically-loadable modules) and C<$link> could be confused with
364 the Unix command line executable of the same name, which does something
365 completely different. Andy Dougherty makes the counter argument "In parrot, I
366 tried to call the command used to link object files and libraries into an
367 executable F<link>, since that's what my vaguely-remembered DOS and VMS
368 experience suggested. I don't think any real confusion has ensued, so it's
369 probably a reasonable name for perl5 to use."
371 "Alas, I've always worried that introducing it would make things worse,
372 since now the module building utilities would have to look for
373 C<$Config{link}> and institute a fall-back plan if it weren't found."
374 Although I can see that as confusing, given that C<$Config{d_link}> is true
375 when (hard) links are available.
377 =head2 Configure Windows using PowerShell
379 Currently, Windows uses hard-coded config files based to build the
380 config.h for compiling Perl. Makefiles are also hard-coded and need to be
381 hand edited prior to building Perl. While this makes it easy to create a perl.exe
382 that works across multiple Windows versions, being able to accurately
383 configure a perl.exe for a specific Windows versions and VS C++ would be
384 a nice enhancement. With PowerShell available on Windows XP and up, this
385 may now be possible. Step 1 might be to investigate whether this is possible
386 and use this to clean up our current makefile situation. Step 2 would be to
387 see if there would be a way to use our existing metaconfig units to configure a
388 Windows Perl or whether we go in a separate direction and make it so. Of
389 course, we all know what step 3 is.
391 =head1 Tasks that need a little C knowledge
393 These tasks would need a little C knowledge, but don't need any specific
394 background or experience with XS, or how the Perl interpreter works
396 =head2 Weed out needless PERL_UNUSED_ARG
398 The C code uses the macro C<PERL_UNUSED_ARG> to stop compilers warning about
399 unused arguments. Often the arguments can't be removed, as there is an
400 external constraint that determines the prototype of the function, so this
401 approach is valid. However, there are some cases where C<PERL_UNUSED_ARG>
402 could be removed. Specifically
408 The prototypes of (nearly all) static functions can be changed
412 Unused arguments generated by short cut macros are wasteful - the short cut
413 macro used can be changed.
417 =head2 Modernize the order of directories in @INC
419 The way @INC is laid out by default, one cannot upgrade core (dual-life)
420 modules without overwriting files. This causes problems for binary
421 package builders. One possible proposal is laid out in this
423 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2002-04/msg02380.html>
427 Natively 64-bit systems need neither -Duse64bitint nor -Duse64bitall.
428 On these systems, it might be the default compilation mode, and there
429 is currently no guarantee that passing no use64bitall option to the
430 Configure process will build a 32bit perl. Implementing -Duse32bit*
431 options would be nice for perl 5.14.
433 =head2 Profile Perl - am I hot or not?
435 The Perl source code is stable enough that it makes sense to profile it,
436 identify and optimise the hotspots. It would be good to measure the
437 performance of the Perl interpreter using free tools such as cachegrind,
438 gprof, and dtrace, and work to reduce the bottlenecks they reveal.
440 As part of this, the idea of F<pp_hot.c> is that it contains the I<hot> ops,
441 the ops that are most commonly used. The idea is that by grouping them, their
442 object code will be adjacent in the executable, so they have a greater chance
443 of already being in the CPU cache (or swapped in) due to being near another op
446 Except that it's not clear if these really are the most commonly used ops. So
447 as part of exercising your skills with coverage and profiling tools you might
448 want to determine what ops I<really> are the most commonly used. And in turn
449 suggest evictions and promotions to achieve a better F<pp_hot.c>.
451 One piece of Perl code that might make a good testbed is F<installman>.
453 =head2 Allocate OPs from arenas
455 Currently all new OP structures are individually malloc()ed and free()d.
456 All C<malloc> implementations have space overheads, and are now as fast as
457 custom allocates so it would both use less memory and less CPU to allocate
458 the various OP structures from arenas. The SV arena code can probably be
461 Note that Configuring perl with C<-Accflags=-DPL_OP_SLAB_ALLOC> will use
462 Perl_Slab_alloc() to pack optrees into a contiguous block, which is
463 probably superior to the use of OP arenas, esp. from a cache locality
464 standpoint. See L<Profile Perl - am I hot or not?>.
466 =head2 Improve win32/wince.c
468 Currently, numerous functions look virtually, if not completely,
469 identical in both F<win32/wince.c> and F<win32/win32.c> files, which can't
472 =head2 Use secure CRT functions when building with VC8 on Win32
474 Visual C++ 2005 (VC++ 8.x) deprecated a number of CRT functions on the basis
475 that they were "unsafe" and introduced differently named secure versions of
476 them as replacements, e.g. instead of writing
478 FILE* f = fopen(__FILE__, "r");
483 errno_t err = fopen_s(&f, __FILE__, "r");
485 Currently, the warnings about these deprecations have been disabled by adding
486 -D_CRT_SECURE_NO_DEPRECATE to the CFLAGS. It would be nice to remove that
487 warning suppressant and actually make use of the new secure CRT functions.
489 There is also a similar issue with POSIX CRT function names like fileno having
490 been deprecated in favour of ISO C++ conformant names like _fileno. These
491 warnings are also currently suppressed by adding -D_CRT_NONSTDC_NO_DEPRECATE. It
492 might be nice to do as Microsoft suggest here too, although, unlike the secure
493 functions issue, there is presumably little or no benefit in this case.
495 =head2 Fix POSIX::access() and chdir() on Win32
497 These functions currently take no account of DACLs and therefore do not behave
498 correctly in situations where access is restricted by DACLs (as opposed to the
499 read-only attribute).
501 Furthermore, POSIX::access() behaves differently for directories having the
502 read-only attribute set depending on what CRT library is being used. For
503 example, the _access() function in the VC6 and VC7 CRTs (wrongly) claim that
504 such directories are not writable, whereas in fact all directories are writable
505 unless access is denied by DACLs. (In the case of directories, the read-only
506 attribute actually only means that the directory cannot be deleted.) This CRT
507 bug is fixed in the VC8 and VC9 CRTs (but, of course, the directory may still
508 not actually be writable if access is indeed denied by DACLs).
510 For the chdir() issue, see ActiveState bug #74552:
511 L<http://bugs.activestate.com/show_bug.cgi?id=74552>
513 Therefore, DACLs should be checked both for consistency across CRTs and for
516 (Note that perl's -w operator should not be modified to check DACLs. It has
517 been written so that it reflects the state of the read-only attribute, even
518 for directories (whatever CRT is being used), for symmetry with chmod().)
520 =head2 strcat(), strcpy(), strncat(), strncpy(), sprintf(), vsprintf()
522 Maybe create a utility that checks after each libperl.a creation that
523 none of the above (nor sprintf(), vsprintf(), or *SHUDDER* gets())
524 ever creep back to libperl.a.
526 nm libperl.a | ./miniperl -alne '$o = $F[0] if /:$/; print "$o $F[1]" if $F[0] eq "U" && $F[1] =~ /^(?:strn?c(?:at|py)|v?sprintf|gets)$/'
528 Note, of course, that this will only tell whether B<your> platform
529 is using those naughty interfaces.
531 =head2 -D_FORTIFY_SOURCE=2, -fstack-protector
533 Recent glibcs support C<-D_FORTIFY_SOURCE=2> and recent gcc
534 (4.1 onwards?) supports C<-fstack-protector>, both of which give
535 protection against various kinds of buffer overflow problems.
536 These should probably be used for compiling Perl whenever available,
537 Configure and/or hints files should be adjusted to probe for the
538 availability of these features and enable them as appropriate.
540 =head2 Arenas for GPs? For MAGIC?
542 C<struct gp> and C<struct magic> are both currently allocated by C<malloc>.
543 It might be a speed or memory saving to change to using arenas. Or it might
544 not. It would need some suitable benchmarking first. In particular, C<GP>s
545 can probably be changed with minimal compatibility impact (probably nothing
546 outside of the core, or even outside of F<gv.c> allocates them), but they
547 probably aren't allocated/deallocated often enough for a speed saving. Whereas
548 C<MAGIC> is allocated/deallocated more often, but in turn, is also something
549 more externally visible, so changing the rules here may bite external code.
553 Several SV body structs are now the same size, notably PVMG and PVGV, PVAV and
554 PVHV, and PVCV and PVFM. It should be possible to allocate and return same
555 sized bodies from the same actual arena, rather than maintaining one arena for
556 each. This could save 4-6K per thread, of memory no longer tied up in the
557 not-yet-allocated part of an arena.
560 =head1 Tasks that need a knowledge of XS
562 These tasks would need C knowledge, and roughly the level of knowledge of
563 the perl API that comes from writing modules that use XS to interface to
566 =head2 Write an XS cookbook
568 Create pod/perlxscookbook.pod with short, task-focused 'recipes' in XS that
569 demonstrate common tasks and good practices. (Some of these might be
570 extracted from perlguts.) The target audience should be XS novices, who need
571 more examples than perlguts but something less overwhelming than perlapi.
572 Recipes should provide "one pretty good way to do it" instead of TIMTOWTDI.
574 Rather than focusing on interfacing Perl to C libraries, such a cookbook
575 should probably focus on how to optimize Perl routines by re-writing them
576 in XS. This will likely be more motivating to those who mostly work in
577 Perl but are looking to take the next step into XS.
579 Deconstructing and explaining some simpler XS modules could be one way to
580 bootstrap a cookbook. (List::Util? Class::XSAccessor? Tree::Ternary_XS?)
581 Another option could be deconstructing the implementation of some simpler
584 =head2 Allow XSUBs to inline themselves as OPs
586 For a simple XSUB, often the subroutine dispatch takes more time than the
587 XSUB itself. The tokeniser already has the ability to inline constant
588 subroutines - it would be good to provide a way to inline other subroutines.
590 Specifically, simplest approach looks to be to allow an XSUB to provide an
591 alternative implementation of itself as a custom OP. A new flag bit in
592 C<CvFLAGS()> would signal to the peephole optimiser to take an optree
595 b <@> leave[1 ref] vKP/REFC ->(end)
597 2 <;> nextstate(main 1 -e:1) v:{ ->3
598 a <2> sassign vKS/2 ->b
599 8 <1> entersub[t2] sKS/TARG,1 ->9
602 4 <$> const(IV 1) sM ->5
603 6 <1> rv2av[t1] lKM/1 ->7
605 - <1> ex-rv2cv sK ->-
606 7 <$> gv(*x) s/EARLYCV ->8
607 - <1> ex-rv2sv sKRM*/1 ->a
610 perform the symbol table lookup of C<rv2cv> and C<gv(*x)>, locate the
611 pointer to the custom OP that provides the direct implementation, and re-
612 write the optree something like:
614 b <@> leave[1 ref] vKP/REFC ->(end)
616 2 <;> nextstate(main 1 -e:1) v:{ ->3
617 a <2> sassign vKS/2 ->b
621 4 <$> const(IV 1) sM ->5
622 6 <1> rv2av[t1] lKM/1 ->7
624 - <1> ex-rv2cv sK ->-
625 - <$> ex-gv(*x) s/EARLYCV ->7
626 - <1> ex-rv2sv sKRM*/1 ->a
629 I<i.e.> the C<gv(*)> OP has been nulled and spliced out of the execution
630 path, and the C<entersub> OP has been replaced by the custom op.
632 This approach should provide a measurable speed up to simple XSUBs inside
633 tight loops. Initially one would have to write the OP alternative
634 implementation by hand, but it's likely that this should be reasonably
635 straightforward for the type of XSUB that would benefit the most. Longer
636 term, once the run-time implementation is proven, it should be possible to
637 progressively update ExtUtils::ParseXS to generate OP implementations for
640 =head2 Remove the use of SVs as temporaries in dump.c
642 F<dump.c> contains debugging routines to dump out the contains of perl data
643 structures, such as C<SV>s, C<AV>s and C<HV>s. Currently, the dumping code
644 B<uses> C<SV>s for its temporary buffers, which was a logical initial
645 implementation choice, as they provide ready made memory handling.
647 However, they also lead to a lot of confusion when it happens that what you're
648 trying to debug is seen by the code in F<dump.c>, correctly or incorrectly, as
649 a temporary scalar it can use for a temporary buffer. It's also not possible
650 to dump scalars before the interpreter is properly set up, such as during
651 ithreads cloning. It would be good to progressively replace the use of scalars
652 as string accumulation buffers with something much simpler, directly allocated
653 by C<malloc>. The F<dump.c> code is (or should be) only producing 7 bit
654 US-ASCII, so output character sets are not an issue.
656 Producing and proving an internal simple buffer allocation would make it easier
657 to re-write the internals of the PerlIO subsystem to avoid using C<SV>s for
658 B<its> buffers, use of which can cause problems similar to those of F<dump.c>,
661 =head2 safely supporting POSIX SA_SIGINFO
663 Some years ago Jarkko supplied patches to provide support for the POSIX
664 SA_SIGINFO feature in Perl, passing the extra data to the Perl signal handler.
666 Unfortunately, it only works with "unsafe" signals, because under safe
667 signals, by the time Perl gets to run the signal handler, the extra
668 information has been lost. Moreover, it's not easy to store it somewhere,
669 as you can't call mutexs, or do anything else fancy, from inside a signal
672 So it strikes me that we could provide safe SA_SIGINFO support
678 Provide global variables for two file descriptors
682 When the first request is made via C<sigaction> for C<SA_SIGINFO>, create a
683 pipe, store the reader in one, the writer in the other
687 In the "safe" signal handler (C<Perl_csighandler()>/C<S_raise_signal()>), if
688 the C<siginfo_t> pointer non-C<NULL>, and the writer file handle is open,
694 serialise signal number, C<struct siginfo_t> (or at least the parts we care
695 about) into a small auto char buff
699 C<write()> that (non-blocking) to the writer fd
705 if it writes 100%, flag the signal in a counter of "signals on the pipe" akin
706 to the current per-signal-number counts
710 if it writes 0%, assume the pipe is full. Flag the data as lost?
714 if it writes partially, croak a panic, as your OS is broken.
722 in the regular C<PERL_ASYNC_CHECK()> processing, if there are "signals on
723 the pipe", read the data out, deserialise, build the Perl structures on
724 the stack (code in C<Perl_sighandler()>, the "unsafe" handler), and call as
729 I think that this gets us decent C<SA_SIGINFO> support, without the current risk
730 of running Perl code inside the signal handler context. (With all the dangers
731 of things like C<malloc> corruption that that currently offers us)
733 For more information see the thread starting with this message:
734 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-03/msg00305.html>
736 =head2 autovivification
738 Make all autovivification consistent w.r.t LVALUE/RVALUE and strict/no strict;
740 This task is incremental - even a little bit of work on it will help.
742 =head2 Unicode in Filenames
744 chdir, chmod, chown, chroot, exec, glob, link, lstat, mkdir, open,
745 opendir, qx, readdir, readlink, rename, rmdir, stat, symlink, sysopen,
746 system, truncate, unlink, utime, -X. All these could potentially accept
747 Unicode filenames either as input or output (and in the case of system
748 and qx Unicode in general, as input or output to/from the shell).
749 Whether a filesystem - an operating system pair understands Unicode in
752 Known combinations that have some level of understanding include
753 Microsoft NTFS, Apple HFS+ (In Mac OS 9 and X) and Apple UFS (in Mac
754 OS X), NFS v4 is rumored to be Unicode, and of course Plan 9. How to
755 create Unicode filenames, what forms of Unicode are accepted and used
756 (UCS-2, UTF-16, UTF-8), what (if any) is the normalization form used,
757 and so on, varies. Finding the right level of interfacing to Perl
758 requires some thought. Remember that an OS does not implicate a
761 (The Windows -C command flag "wide API support" has been at least
762 temporarily retired in 5.8.1, and the -C has been repurposed, see
765 Most probably the right way to do this would be this:
766 L</"Virtualize operating system access">.
768 =head2 Unicode in %ENV
770 Currently the %ENV entries are always byte strings.
771 See L</"Virtualize operating system access">.
773 =head2 Unicode and glob()
775 Currently glob patterns and filenames returned from File::Glob::glob()
776 are always byte strings. See L</"Virtualize operating system access">.
778 =head2 use less 'memory'
780 Investigate trade offs to switch out perl's choices on memory usage.
781 Particularly perl should be able to give memory back.
783 This task is incremental - even a little bit of work on it will help.
785 =head2 Re-implement C<:unique> in a way that is actually thread-safe
787 The old implementation made bad assumptions on several levels. A good 90%
788 solution might be just to make C<:unique> work to share the string buffer
789 of SvPVs. That way large constant strings can be shared between ithreads,
790 such as the configuration information in F<Config>.
792 =head2 Make tainting consistent
794 Tainting would be easier to use if it didn't take documented shortcuts and
795 allow taint to "leak" everywhere within an expression.
797 =head2 readpipe(LIST)
799 system() accepts a LIST syntax (and a PROGRAM LIST syntax) to avoid
800 running a shell. readpipe() (the function behind qx//) could be similarly
803 =head2 Audit the code for destruction ordering assumptions
807 /* Need to check SvMAGICAL, as during global destruction it may be that
808 AvARYLEN(av) has been freed before av, and hence the SvANY() pointer
809 is now part of the linked list of SV heads, rather than pointing to
810 the original body. */
811 /* FIXME - audit the code for other bugs like this one. */
813 adding the C<SvMAGICAL> check to
815 if (AvARYLEN(av) && SvMAGICAL(AvARYLEN(av))) {
816 MAGIC *mg = mg_find (AvARYLEN(av), PERL_MAGIC_arylen);
818 Go through the core and look for similar assumptions that SVs have particular
819 types, as all bets are off during global destruction.
821 =head2 Extend PerlIO and PerlIO::Scalar
823 PerlIO::Scalar doesn't know how to truncate(). Implementing this
824 would require extending the PerlIO vtable.
826 Similarly the PerlIO vtable doesn't know about formats (write()), or
827 about stat(), or chmod()/chown(), utime(), or flock().
829 (For PerlIO::Scalar it's hard to see what e.g. mode bits or ownership
832 PerlIO doesn't do directories or symlinks, either: mkdir(), rmdir(),
833 opendir(), closedir(), seekdir(), rewinddir(), glob(); symlink(),
836 See also L</"Virtualize operating system access">.
838 =head2 -C on the #! line
840 It should be possible to make -C work correctly if found on the #! line,
841 given that all perl command line options are strict ASCII, and -C changes
842 only the interpretation of non-ASCII characters, and not for the script file
843 handle. To make it work needs some investigation of the ordering of function
844 calls during startup, and (by implication) a bit of tweaking of that order.
846 =head2 Organize error messages
848 Perl's diagnostics (error messages, see L<perldiag>) could use
849 reorganizing and formalizing so that each error message has its
850 stable-for-all-eternity unique id, categorized by severity, type, and
851 subsystem. (The error messages would be listed in a datafile outside
852 of the Perl source code, and the source code would only refer to the
853 messages by the id.) This clean-up and regularizing should apply
854 for all croak() messages.
856 This would enable all sorts of things: easier translation/localization
857 of the messages (though please do keep in mind the caveats of
858 L<Locale::Maketext> about too straightforward approaches to
859 translation), filtering by severity, and instead of grepping for a
860 particular error message one could look for a stable error id. (Of
861 course, changing the error messages by default would break all the
862 existing software depending on some particular error message...)
864 This kind of functionality is known as I<message catalogs>. Look for
865 inspiration for example in the catgets() system, possibly even use it
866 if available-- but B<only> if available, all platforms will B<not>
869 For the really pure at heart, consider extending this item to cover
870 also the warning messages (see L<perllexwarn>, C<warnings.pl>).
872 =head1 Tasks that need a knowledge of the interpreter
874 These tasks would need C knowledge, and knowledge of how the interpreter works,
875 or a willingness to learn.
877 =head2 forbid labels with keyword names
879 Currently C<goto keyword> "computes" the label value:
881 $ perl -e 'goto print'
882 Can't find label 1 at -e line 1.
884 It is controversial if the right way to avoid the confusion is to forbid
885 labels with keyword names, or if it would be better to always treat
886 bareword expressions after a "goto" as a label and never as a keyword.
888 =head2 truncate() prototype
890 The prototype of truncate() is currently C<$$>. It should probably
891 be C<*$> instead. (This is changed in F<opcode.pl>)
893 =head2 decapsulation of smart match argument
895 Currently C<$foo ~~ $object> will die with the message "Smart matching a
896 non-overloaded object breaks encapsulation". It would be nice to allow
897 to bypass this by using explicitly the syntax C<$foo ~~ %$object> or
900 =head2 error reporting of [$a ; $b]
902 Using C<;> inside brackets is a syntax error, and we don't propose to change
903 that by giving it any meaning. However, it's not reported very helpfully:
905 $ perl -e '$a = [$b; $c];'
906 syntax error at -e line 1, near "$b;"
907 syntax error at -e line 1, near "$c]"
908 Execution of -e aborted due to compilation errors.
910 It should be possible to hook into the tokeniser or the lexer, so that when a
911 C<;> is parsed where it is not legal as a statement terminator (ie inside
912 C<{}> used as a hashref, C<[]> or C<()>) it issues an error something like
913 I<';' isn't legal inside an expression - if you need multiple statements use a
914 do {...} block>. See the thread starting at
915 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-09/msg00573.html>
917 =head2 lexicals used only once
921 $ perl -we '$pie = 42'
922 Name "main::pie" used only once: possible typo at -e line 1.
926 $ perl -we 'my $pie = 42'
928 Logically all lexicals used only once should warn, if the user asks for
929 warnings. An unworked RT ticket (#5087) has been open for almost seven
930 years for this discrepancy.
934 The handling of Unicode is unclean in many places. In the regex engine
935 there are especially many problems. The swash data structure could be
936 replaced my something better. Inversion lists and maps are likely
937 candidates. The whole Unicode database could be placed in-core for a
938 huge speed-up. Only minimal work was done on the optimizer when utf8
939 was added, with the result that the synthetic start class often will
940 fail to narrow down the possible choices when given non-Latin1 input.
942 =head2 Properly Unicode safe tokeniser and pads.
944 The tokeniser isn't actually very UTF-8 clean. C<use utf8;> is a hack -
945 variable names are stored in stashes as raw bytes, without the utf-8 flag
946 set. The pad API only takes a C<char *> pointer, so that's all bytes too. The
947 tokeniser ignores the UTF-8-ness of C<PL_rsfp>, or any SVs returned from
948 source filters. All this could be fixed.
950 =head2 state variable initialization in list context
952 Currently this is illegal:
954 state ($a, $b) = foo();
956 In Perl 6, C<state ($a) = foo();> and C<(state $a) = foo();> have different
957 semantics, which is tricky to implement in Perl 5 as currently they produce
958 the same opcode trees. The Perl 6 design is firm, so it would be good to
959 implement the necessary code in Perl 5. There are comments in
960 C<Perl_newASSIGNOP()> that show the code paths taken by various assignment
961 constructions involving state variables.
963 =head2 Implement $value ~~ 0 .. $range
965 It would be nice to extend the syntax of the C<~~> operator to also
966 understand numeric (and maybe alphanumeric) ranges.
968 =head2 A does() built-in
970 Like ref(), only useful. It would call the C<DOES> method on objects; it
971 would also tell whether something can be dereferenced as an
972 array/hash/etc., or used as a regexp, etc.
973 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-03/msg00481.html>
975 =head2 Tied filehandles and write() don't mix
977 There is no method on tied filehandles to allow them to be called back by
980 =head2 Propagate compilation hints to the debugger
982 Currently a debugger started with -dE on the command-line doesn't see the
983 features enabled by -E. More generally hints (C<$^H> and C<%^H>) aren't
984 propagated to the debugger. Probably it would be a good thing to propagate
985 hints from the innermost non-C<DB::> scope: this would make code eval'ed
986 in the debugger see the features (and strictures, etc.) currently in
989 =head2 Attach/detach debugger from running program
991 The old perltodo notes "With C<gdb>, you can attach the debugger to a running
992 program if you pass the process ID. It would be good to do this with the Perl
993 debugger on a running Perl program, although I'm not sure how it would be
994 done." ssh and screen do this with named pipes in /tmp. Maybe we can too.
996 =head2 LVALUE functions for lists
998 The old perltodo notes that lvalue functions don't work for list or hash
999 slices. This would be good to fix.
1001 =head2 regexp optimiser optional
1003 The regexp optimiser is not optional. It should configurable to be, to allow
1004 its performance to be measured, and its bugs to be easily demonstrated.
1006 =head2 C</w> regex modifier
1008 That flag would enable to match whole words, and also to interpolate
1009 arrays as alternations. With it, C</P/w> would be roughly equivalent to:
1011 do { local $"='|'; /\b(?:P)\b/ }
1014 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-01/msg00400.html>
1017 =head2 optional optimizer
1019 Make the peephole optimizer optional. Currently it performs two tasks as
1020 it walks the optree - genuine peephole optimisations, and necessary fixups of
1021 ops. It would be good to find an efficient way to switch out the
1022 optimisations whilst keeping the fixups.
1024 =head2 You WANT *how* many
1026 Currently contexts are void, scalar and list. split has a special mechanism in
1027 place to pass in the number of return values wanted. It would be useful to
1028 have a general mechanism for this, backwards compatible and little speed hit.
1029 This would allow proposals such as short circuiting sort to be implemented
1030 as a module on CPAN.
1032 =head2 lexical aliases
1034 Allow lexical aliases (maybe via the syntax C<my \$alias = \$foo>.
1036 =head2 entersub XS vs Perl
1038 At the moment pp_entersub is huge, and has code to deal with entering both
1039 perl and XS subroutines. Subroutine implementations rarely change between
1040 perl and XS at run time, so investigate using 2 ops to enter subs (one for
1041 XS, one for perl) and swap between if a sub is redefined.
1045 Self-ties are currently illegal because they caused too many segfaults. Maybe
1046 the causes of these could be tracked down and self-ties on all types
1049 =head2 Optimize away @_
1051 The old perltodo notes "Look at the "reification" code in C<av.c>".
1053 =head2 Virtualize operating system access
1055 Implement a set of "vtables" that virtualizes operating system access
1056 (open(), mkdir(), unlink(), readdir(), getenv(), etc.) At the very
1057 least these interfaces should take SVs as "name" arguments instead of
1058 bare char pointers; probably the most flexible and extensible way
1059 would be for the Perl-facing interfaces to accept HVs. The system
1060 needs to be per-operating-system and per-file-system
1061 hookable/filterable, preferably both from XS and Perl level
1062 (L<perlport/"Files and Filesystems"> is good reading at this point,
1063 in fact, all of L<perlport> is.)
1065 This has actually already been implemented (but only for Win32),
1066 take a look at F<iperlsys.h> and F<win32/perlhost.h>. While all Win32
1067 variants go through a set of "vtables" for operating system access,
1068 non-Win32 systems currently go straight for the POSIX/Unix-style
1069 system/library call. Similar system as for Win32 should be
1070 implemented for all platforms. The existing Win32 implementation
1071 probably does not need to survive alongside this proposed new
1072 implementation, the approaches could be merged.
1074 What would this give us? One often-asked-for feature this would
1075 enable is using Unicode for filenames, and other "names" like %ENV,
1076 usernames, hostnames, and so forth.
1077 (See L<perlunicode/"When Unicode Does Not Happen">.)
1079 But this kind of virtualization would also allow for things like
1080 virtual filesystems, virtual networks, and "sandboxes" (though as long
1081 as dynamic loading of random object code is allowed, not very safe
1082 sandboxes since external code of course know not of Perl's vtables).
1083 An example of a smaller "sandbox" is that this feature can be used to
1084 implement per-thread working directories: Win32 already does this.
1086 See also L</"Extend PerlIO and PerlIO::Scalar">.
1088 =head2 Investigate PADTMP hash pessimisation
1090 The peephole optimiser converts constants used for hash key lookups to shared
1091 hash key scalars. Under ithreads, something is undoing this work.
1093 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-09/msg00793.html>
1095 =head2 Store the current pad in the OP slab allocator
1098 I hope that I got that "current pad" part correct
1100 Currently we leak ops in various cases of parse failure. I suggested that we
1101 could solve this by always using the op slab allocator, and walking it to
1102 free ops. Dave comments that as some ops are already freed during optree
1103 creation one would have to mark which ops are freed, and not double free them
1104 when walking the slab. He notes that one problem with this is that for some ops
1105 you have to know which pad was current at the time of allocation, which does
1106 change. I suggested storing a pointer to the current pad in the memory allocated
1107 for the slab, and swapping to a new slab each time the pad changes. Dave thinks
1108 that this would work.
1110 =head2 repack the optree
1112 Repacking the optree after execution order is determined could allow
1113 removal of NULL ops, and optimal ordering of OPs with respect to cache-line
1114 filling. The slab allocator could be reused for this purpose. I think that
1115 the best way to do this is to make it an optional step just before the
1116 completed optree is attached to anything else, and to use the slab allocator
1117 unchanged, so that freeing ops is identical whether or not this step runs.
1118 Note that the slab allocator allocates ops downwards in memory, so one would
1119 have to actually "allocate" the ops in reverse-execution order to get them
1120 contiguous in memory in execution order.
1123 L<http://www.nntp.perl.org/group/perl.perl5.porters/2007/12/msg131975.html>
1125 Note that running this copy, and then freeing all the old location ops would
1126 cause their slabs to be freed, which would eliminate possible memory wastage if
1127 the previous suggestion is implemented, and we swap slabs more frequently.
1129 =head2 eliminate incorrect line numbers in warnings
1137 } elsif ($undef == 0) {
1140 used to produce this output:
1142 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1143 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1145 where the line of the second warning was misreported - it should be line 5.
1146 Rafael fixed this - the problem arose because there was no nextstate OP
1147 between the execution of the C<if> and the C<elsif>, hence C<PL_curcop> still
1148 reports that the currently executing line is line 4. The solution was to inject
1149 a nextstate OPs for each C<elsif>, although it turned out that the nextstate
1150 OP needed to be a nulled OP, rather than a live nextstate OP, else other line
1151 numbers became misreported. (Jenga!)
1153 The problem is more general than C<elsif> (although the C<elsif> case is the
1154 most common and the most confusing). Ideally this code
1164 would produce this output
1166 Use of uninitialized value $undef in addition (+) at wrong.pl line 4.
1167 Use of uninitialized value $undef in addition (+) at wrong.pl line 7.
1169 (rather than lines 4 and 5), but this would seem to require every OP to carry
1170 (at least) line number information.
1172 What might work is to have an optional line number in memory just before the
1173 BASEOP structure, with a flag bit in the op to say whether it's present.
1174 Initially during compile every OP would carry its line number. Then add a late
1175 pass to the optimiser (potentially combined with L</repack the optree>) which
1176 looks at the two ops on every edge of the graph of the execution path. If
1177 the line number changes, flags the destination OP with this information.
1178 Once all paths are traced, replace every op with the flag with a
1179 nextstate-light op (that just updates C<PL_curcop>), which in turn then passes
1180 control on to the true op. All ops would then be replaced by variants that
1181 do not store the line number. (Which, logically, why it would work best in
1182 conjunction with L</repack the optree>, as that is already copying/reallocating
1185 (Although I should note that we're not certain that doing this for the general
1188 =head2 optimize tail-calls
1190 Tail-calls present an opportunity for broadly applicable optimization;
1191 anywhere that C<< return foo(...) >> is called, the outer return can
1192 be replaced by a goto, and foo will return directly to the outer
1193 caller, saving (conservatively) 25% of perl's call&return cost, which
1194 is relatively higher than in C. The scheme language is known to do
1195 this heavily. B::Concise provides good insight into where this
1196 optimization is possible, ie anywhere entersub,leavesub op-sequence
1199 perl -MO=Concise,-exec,a,b,-main -e 'sub a{ 1 }; sub b {a()}; b(2)'
1201 Bottom line on this is probably a new pp_tailcall function which
1202 combines the code in pp_entersub, pp_leavesub. This should probably
1203 be done 1st in XS, and using B::Generate to patch the new OP into the
1206 =head2 Add C<00dddd>
1208 It has been proposed that octal constants be specifiable through the syntax
1209 C<0oddddd>, parallel to the existing construct to specify hex constants
1214 Tasks that will get your name mentioned in the description of the "Highlights
1217 =head2 make ithreads more robust
1219 Generally make ithreads more robust. See also L</iCOW>
1221 This task is incremental - even a little bit of work on it will help, and
1222 will be greatly appreciated.
1224 One bit would be to determine how to clone directory handles on systems
1225 without a C<fchdir> function (in sv.c:Perl_dirp_dup).
1227 Fix Perl_sv_dup, et al so that threads can return objects.
1231 Sarathy and Arthur have a proposal for an improved Copy On Write which
1232 specifically will be able to COW new ithreads. If this can be implemented
1233 it would be a good thing.
1235 =head2 (?{...}) closures in regexps
1237 Fix (or rewrite) the implementation of the C</(?{...})/> closures.
1239 =head2 Add class set operations to regexp engine
1241 Apparently these are quite useful. Anyway, Jeffery Friedl wants them.
1243 demerphq has this on his todo list, but right at the bottom.
1246 =head1 Tasks for microperl
1249 [ Each and every one of these may be obsolete, but they were listed
1250 in the old Todo.micro file]
1253 =head2 make creating uconfig.sh automatic
1255 =head2 make creating Makefile.micro automatic
1257 =head2 do away with fork/exec/wait?
1259 (system, popen should be enough?)
1261 =head2 some of the uconfig.sh really needs to be probed (using cc) in buildtime:
1263 (uConfigure? :-) native datatype widths and endianness come to mind