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 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:
20 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/
22 What can we offer you in return? Fame, fortune, and everlasting glory? Maybe
23 not, but if your patch is incorporated, then we'll add your name to the
24 F<AUTHORS> file, which ships in the official distribution. How many other
25 programming languages offer you 1 line of immortality?
27 =head1 Tasks that only need Perl knowledge
29 =head2 Migrate t/ from custom TAP generation
31 Many tests below F<t/> still generate TAP by "hand", rather than using library
32 functions. As explained in L<perlhack/Writing a test>, tests in F<t/> are
33 written in a particular way to test that more complex constructions actually
34 work before using them routinely. Hence they don't use C<Test::More>, but
35 instead there is an intentionally simpler library, F<t/test.pl>. However,
36 quite a few tests in F<t/> have not been refactored to use it. Refactoring
37 any of these tests, one at a time, is a useful thing TODO.
39 The subdirectories F<base>, F<cmd> and F<comp>, that contain the most
40 basic tests, should be excluded from this task.
42 =head2 Automate perldelta generation
44 The perldelta file accompanying each release summaries the major changes.
45 It's mostly manually generated currently, but some of that could be
46 automated with a bit of perl, specifically the generation of
50 =item Modules and Pragmata
52 =item New Documentation
58 See F<Porting/how_to_write_a_perldelta.pod> for details.
60 =head2 Remove duplication of test setup.
62 Schwern notes, that there's duplication of code - lots and lots of tests have
63 some variation on the big block of C<$Is_Foo> checks. We can safely put this
64 into a file, change it to build an C<%Is> hash and require it. Maybe just put
65 it into F<test.pl>. Throw in the handy tainting subroutines.
67 =head2 POD -E<gt> HTML conversion in the core still sucks
69 Which is crazy given just how simple POD purports to be, and how simple HTML
70 can be. It's not actually I<as> simple as it sounds, particularly with the
71 flexibility POD allows for C<=item>, but it would be good to improve the
72 visual appeal of the HTML generated, and to avoid it having any validation
73 errors. See also L</make HTML install work>, as the layout of installation tree
74 is needed to improve the cross-linking.
76 The addition of C<Pod::Simple> and its related modules may make this task
79 =head2 Make ExtUtils::ParseXS use strict;
81 F<lib/ExtUtils/ParseXS.pm> contains this line
83 # use strict; # One of these days...
85 Simply uncomment it, and fix all the resulting issues :-)
87 The more practical approach, to break the task down into manageable chunks, is
88 to work your way though the code from bottom to top, or if necessary adding
89 extra C<{ ... }> blocks, and turning on strict within them.
91 =head2 Make Schwern poorer
93 We should have tests for everything. When all the core's modules are tested,
94 Schwern has promised to donate to $500 to TPF. We may need volunteers to
95 hold him upside down and shake vigorously in order to actually extract the
98 =head2 Improve the coverage of the core tests
100 Use Devel::Cover to ascertain the core modules' test coverage, then add
101 tests that are currently missing.
105 A full test suite for the B module would be nice.
107 =head2 A decent benchmark
109 C<perlbench> seems impervious to any recent changes made to the perl core. It
110 would be useful to have a reasonable general benchmarking suite that roughly
111 represented what current perl programs do, and measurably reported whether
112 tweaks to the core improve, degrade or don't really affect performance, to
113 guide people attempting to optimise the guts of perl. Gisle would welcome
114 new tests for perlbench.
116 =head2 fix tainting bugs
118 Fix the bugs revealed by running the test suite with the C<-t> switch (via
119 C<make test.taintwarn>).
121 =head2 Dual life everything
123 As part of the "dists" plan, anything that doesn't belong in the smallest perl
124 distribution needs to be dual lifed. Anything else can be too. Figure out what
125 changes would be needed to package that module and its tests up for CPAN, and
126 do so. Test it with older perl releases, and fix the problems you find.
128 To make a minimal perl distribution, it's useful to look at
129 F<t/lib/commonsense.t>.
131 =head2 POSIX memory footprint
133 Ilya observed that use POSIX; eats memory like there's no tomorrow, and at
134 various times worked to cut it down. There is probably still fat to cut out -
135 for example POSIX passes Exporter some very memory hungry data structures.
137 =head2 embed.pl/makedef.pl
139 There is a script F<embed.pl> that generates several header files to prefix
140 all of Perl's symbols in a consistent way, to provide some semblance of
141 namespace support in C<C>. Functions are declared in F<embed.fnc>, variables
142 in F<interpvar.h>. Quite a few of the functions and variables
143 are conditionally declared there, using C<#ifdef>. However, F<embed.pl>
144 doesn't understand the C macros, so the rules about which symbols are present
145 when is duplicated in F<makedef.pl>. Writing things twice is bad, m'kay.
146 It would be good to teach C<embed.pl> to understand the conditional
147 compilation, and hence remove the duplication, and the mistakes it has caused.
149 =head2 use strict; and AutoLoad
151 Currently if you write
154 use AutoLoader 'AUTOLOAD';
159 print join (' ', No, strict, here), "!\n";
162 then C<use strict;> isn't in force within the autoloaded subroutines. It would
163 be more consistent (and less surprising) to arrange for all lexical pragmas
164 in force at the __END__ block to be in force within each autoloaded subroutine.
166 There's a similar problem with SelfLoader.
168 =head2 profile installman
170 The F<installman> script is slow. All it is doing text processing, which we're
171 told is something Perl is good at. So it would be nice to know what it is doing
172 that is taking so much CPU, and where possible address it.
174 =head2 enable lexical enabling/disabling of individual warnings
176 Currently, warnings can only be enabled or disabled by category. There
177 are times when it would be useful to quash a single warning, not a
180 =head1 Tasks that need a little sysadmin-type knowledge
182 Or if you prefer, tasks that you would learn from, and broaden your skills
185 =head2 make HTML install work
187 There is an C<installhtml> target in the Makefile. It's marked as
188 "experimental". It would be good to get this tested, make it work reliably, and
189 remove the "experimental" tag. This would include
195 Checking that cross linking between various parts of the documentation works.
196 In particular that links work between the modules (files with POD in F<lib/>)
197 and the core documentation (files in F<pod/>)
201 Work out how to split C<perlfunc> into chunks, preferably one per function
202 group, preferably with general case code that could be used elsewhere.
203 Challenges here are correctly identifying the groups of functions that go
204 together, and making the right named external cross-links point to the right
205 page. Things to be aware of are C<-X>, groups such as C<getpwnam> to
206 C<endservent>, two or more C<=items> giving the different parameter lists, such
209 =item substr EXPR,OFFSET,LENGTH,REPLACEMENT
210 =item substr EXPR,OFFSET,LENGTH
211 =item substr EXPR,OFFSET
213 and different parameter lists having different meanings. (eg C<select>)
217 =head2 compressed man pages
219 Be able to install them. This would probably need a configure test to see how
220 the system does compressed man pages (same directory/different directory?
221 same filename/different filename), as well as tweaking the F<installman> script
222 to compress as necessary.
224 =head2 Add a code coverage target to the Makefile
226 Make it easy for anyone to run Devel::Cover on the core's tests. The steps
227 to do this manually are roughly
233 do a normal C<Configure>, but include Devel::Cover as a module to install
234 (see L<INSTALL> for how to do this)
242 cd t; HARNESS_PERL_SWITCHES=-MDevel::Cover ./perl -I../lib harness
246 Process the resulting Devel::Cover database
250 This just give you the coverage of the F<.pm>s. To also get the C level
257 Additionally tell C<Configure> to use the appropriate C compiler flags for
264 (instead of C<make perl>)
268 After running the tests run C<gcov> to generate all the F<.gcov> files.
269 (Including down in the subdirectories of F<ext/>
273 (From the top level perl directory) run C<gcov2perl> on all the C<.gcov> files
274 to get their stats into the cover_db directory.
278 Then process the Devel::Cover database
282 It would be good to add a single switch to C<Configure> to specify that you
283 wanted to perform perl level coverage, and another to specify C level
284 coverage, and have C<Configure> and the F<Makefile> do all the right things
287 =head2 Make Config.pm cope with differences between built and installed perl
289 Quite often vendors ship a perl binary compiled with their (pay-for)
290 compilers. People install a free compiler, such as gcc. To work out how to
291 build extensions, Perl interrogates C<%Config>, so in this situation
292 C<%Config> describes compilers that aren't there, and extension building
293 fails. This forces people into choosing between re-compiling perl themselves
294 using the compiler they have, or only using modules that the vendor ships.
296 It would be good to find a way teach C<Config.pm> about the installation setup,
297 possibly involving probing at install time or later, so that the C<%Config> in
298 a binary distribution better describes the installed machine, when the
299 installed machine differs from the build machine in some significant way.
301 =head2 linker specification files
303 Some platforms mandate that you provide a list of a shared library's external
304 symbols to the linker, so the core already has the infrastructure in place to
305 do this for generating shared perl libraries. My understanding is that the
306 GNU toolchain can accept an optional linker specification file, and restrict
307 visibility just to symbols declared in that file. It would be good to extend
308 F<makedef.pl> to support this format, and to provide a means within
309 C<Configure> to enable it. This would allow Unix users to test that the
310 export list is correct, and to build a perl that does not pollute the global
311 namespace with private symbols, and will fail in the same way as msvc or mingw
312 builds or when using PERL_DL_NONLAZY=1.
314 =head2 Cross-compile support
316 Currently C<Configure> understands C<-Dusecrosscompile> option. This option
317 arranges for building C<miniperl> for TARGET machine, so this C<miniperl> is
318 assumed then to be copied to TARGET machine and used as a replacement of full
321 This could be done little differently. Namely C<miniperl> should be built for
322 HOST and then full C<perl> with extensions should be compiled for TARGET.
323 This, however, might require extra trickery for %Config: we have one config
324 first for HOST and then another for TARGET. Tools like MakeMaker will be
325 mightily confused. Having around two different types of executables and
326 libraries (HOST and TARGET) makes life interesting for Makefiles and
327 shell (and Perl) scripts. There is $Config{run}, normally empty, which
328 can be used as an execution wrapper. Also note that in some
329 cross-compilation/execution environments the HOST and the TARGET do
330 not see the same filesystem(s), the $Config{run} may need to do some
331 file/directory copying back and forth.
335 Make F<pod/roffitall> be updated by F<pod/buildtoc>.
337 =head2 Split "linker" from "compiler"
339 Right now, Configure probes for two commands, and sets two variables:
343 =item * C<cc> (in F<cc.U>)
345 This variable holds the name of a command to execute a C compiler which
346 can resolve multiple global references that happen to have the same
347 name. Usual values are F<cc> and F<gcc>.
348 Fervent ANSI compilers may be called F<c89>. AIX has F<xlc>.
350 =item * C<ld> (in F<dlsrc.U>)
352 This variable indicates the program to be used to link
353 libraries for dynamic loading. On some systems, it is F<ld>.
354 On ELF systems, it should be C<$cc>. Mostly, we'll try to respect
355 the hint file setting.
359 There is an implicit historical assumption from around Perl5.000alpha
360 something, that C<$cc> is also the correct command for linking object files
361 together to make an executable. This may be true on Unix, but it's not true
362 on other platforms, and there are a maze of work arounds in other places (such
363 as F<Makefile.SH>) to cope with this.
365 Ideally, we should create a new variable to hold the name of the executable
366 linker program, probe for it in F<Configure>, and centralise all the special
367 case logic there or in hints files.
369 A small bikeshed issue remains - what to call it, given that C<$ld> is already
370 taken (arguably for the wrong thing now, but on SunOS 4.1 it is the command
371 for creating dynamically-loadable modules) and C<$link> could be confused with
372 the Unix command line executable of the same name, which does something
373 completely different. Andy Dougherty makes the counter argument "In parrot, I
374 tried to call the command used to link object files and libraries into an
375 executable F<link>, since that's what my vaguely-remembered DOS and VMS
376 experience suggested. I don't think any real confusion has ensued, so it's
377 probably a reasonable name for perl5 to use."
379 "Alas, I've always worried that introducing it would make things worse,
380 since now the module building utilities would have to look for
381 C<$Config{link}> and institute a fall-back plan if it weren't found."
382 Although I can see that as confusing, given that C<$Config{d_link}> is true
383 when (hard) links are available.
385 =head2 Configure Windows using PowerShell
387 Currently, Windows uses hard-coded config files based to build the
388 config.h for compiling Perl. Makefiles are also hard-coded and need to be
389 hand edited prior to building Perl. While this makes it easy to create a perl.exe
390 that works across multiple Windows versions, being able to accurately
391 configure a perl.exe for a specific Windows versions and VS C++ would be
392 a nice enhancement. With PowerShell available on Windows XP and up, this
393 may now be possible. Step 1 might be to investigate whether this is possible
394 and use this to clean up our current makefile situation. Step 2 would be to
395 see if there would be a way to use our existing metaconfig units to configure a
396 Windows Perl or whether we go in a separate direction and make it so. Of
397 course, we all know what step 3 is.
399 =head2 decouple -g and -DDEBUGGING
401 Currently F<Configure> automatically adds C<-DDEBUGGING> to the C compiler
402 flags if it spots C<-g> in the optimiser flags. The pre-processor directive
403 C<DEBUGGING> enables F<perl>'s command line C<-D> options, but in the process
404 makes F<perl> slower. It would be good to disentangle this logic, so that
405 C-level debugging with C<-g> and Perl level debugging with C<-D> can easily
406 be enabled independently.
408 =head1 Tasks that need a little C knowledge
410 These tasks would need a little C knowledge, but don't need any specific
411 background or experience with XS, or how the Perl interpreter works
413 =head2 Weed out needless PERL_UNUSED_ARG
415 The C code uses the macro C<PERL_UNUSED_ARG> to stop compilers warning about
416 unused arguments. Often the arguments can't be removed, as there is an
417 external constraint that determines the prototype of the function, so this
418 approach is valid. However, there are some cases where C<PERL_UNUSED_ARG>
419 could be removed. Specifically
425 The prototypes of (nearly all) static functions can be changed
429 Unused arguments generated by short cut macros are wasteful - the short cut
430 macro used can be changed.
434 =head2 Modernize the order of directories in @INC
436 The way @INC is laid out by default, one cannot upgrade core (dual-life)
437 modules without overwriting files. This causes problems for binary
438 package builders. One possible proposal is laid out in this
440 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2002-04/msg02380.html>.
444 Natively 64-bit systems need neither -Duse64bitint nor -Duse64bitall.
445 On these systems, it might be the default compilation mode, and there
446 is currently no guarantee that passing no use64bitall option to the
447 Configure process will build a 32bit perl. Implementing -Duse32bit*
448 options would be nice for perl 5.14.
450 =head2 Profile Perl - am I hot or not?
452 The Perl source code is stable enough that it makes sense to profile it,
453 identify and optimise the hotspots. It would be good to measure the
454 performance of the Perl interpreter using free tools such as cachegrind,
455 gprof, and dtrace, and work to reduce the bottlenecks they reveal.
457 As part of this, the idea of F<pp_hot.c> is that it contains the I<hot> ops,
458 the ops that are most commonly used. The idea is that by grouping them, their
459 object code will be adjacent in the executable, so they have a greater chance
460 of already being in the CPU cache (or swapped in) due to being near another op
463 Except that it's not clear if these really are the most commonly used ops. So
464 as part of exercising your skills with coverage and profiling tools you might
465 want to determine what ops I<really> are the most commonly used. And in turn
466 suggest evictions and promotions to achieve a better F<pp_hot.c>.
468 One piece of Perl code that might make a good testbed is F<installman>.
470 =head2 Allocate OPs from arenas
472 Currently all new OP structures are individually malloc()ed and free()d.
473 All C<malloc> implementations have space overheads, and are now as fast as
474 custom allocates so it would both use less memory and less CPU to allocate
475 the various OP structures from arenas. The SV arena code can probably be
478 Note that Configuring perl with C<-Accflags=-DPL_OP_SLAB_ALLOC> will use
479 Perl_Slab_alloc() to pack optrees into a contiguous block, which is
480 probably superior to the use of OP arenas, esp. from a cache locality
481 standpoint. See L<Profile Perl - am I hot or not?>.
483 =head2 Improve win32/wince.c
485 Currently, numerous functions look virtually, if not completely,
486 identical in both C<win32/wince.c> and C<win32/win32.c> files, which can't
489 =head2 Use secure CRT functions when building with VC8 on Win32
491 Visual C++ 2005 (VC++ 8.x) deprecated a number of CRT functions on the basis
492 that they were "unsafe" and introduced differently named secure versions of
493 them as replacements, e.g. instead of writing
495 FILE* f = fopen(__FILE__, "r");
500 errno_t err = fopen_s(&f, __FILE__, "r");
502 Currently, the warnings about these deprecations have been disabled by adding
503 -D_CRT_SECURE_NO_DEPRECATE to the CFLAGS. It would be nice to remove that
504 warning suppressant and actually make use of the new secure CRT functions.
506 There is also a similar issue with POSIX CRT function names like fileno having
507 been deprecated in favour of ISO C++ conformant names like _fileno. These
508 warnings are also currently suppressed by adding -D_CRT_NONSTDC_NO_DEPRECATE. It
509 might be nice to do as Microsoft suggest here too, although, unlike the secure
510 functions issue, there is presumably little or no benefit in this case.
512 =head2 Fix POSIX::access() and chdir() on Win32
514 These functions currently take no account of DACLs and therefore do not behave
515 correctly in situations where access is restricted by DACLs (as opposed to the
516 read-only attribute).
518 Furthermore, POSIX::access() behaves differently for directories having the
519 read-only attribute set depending on what CRT library is being used. For
520 example, the _access() function in the VC6 and VC7 CRTs (wrongly) claim that
521 such directories are not writable, whereas in fact all directories are writable
522 unless access is denied by DACLs. (In the case of directories, the read-only
523 attribute actually only means that the directory cannot be deleted.) This CRT
524 bug is fixed in the VC8 and VC9 CRTs (but, of course, the directory may still
525 not actually be writable if access is indeed denied by DACLs).
527 For the chdir() issue, see ActiveState bug #74552:
528 http://bugs.activestate.com/show_bug.cgi?id=74552
530 Therefore, DACLs should be checked both for consistency across CRTs and for
533 (Note that perl's -w operator should not be modified to check DACLs. It has
534 been written so that it reflects the state of the read-only attribute, even
535 for directories (whatever CRT is being used), for symmetry with chmod().)
537 =head2 strcat(), strcpy(), strncat(), strncpy(), sprintf(), vsprintf()
539 Maybe create a utility that checks after each libperl.a creation that
540 none of the above (nor sprintf(), vsprintf(), or *SHUDDER* gets())
541 ever creep back to libperl.a.
543 nm libperl.a | ./miniperl -alne '$o = $F[0] if /:$/; print "$o $F[1]" if $F[0] eq "U" && $F[1] =~ /^(?:strn?c(?:at|py)|v?sprintf|gets)$/'
545 Note, of course, that this will only tell whether B<your> platform
546 is using those naughty interfaces.
548 =head2 -D_FORTIFY_SOURCE=2, -fstack-protector
550 Recent glibcs support C<-D_FORTIFY_SOURCE=2> and recent gcc
551 (4.1 onwards?) supports C<-fstack-protector>, both of which give
552 protection against various kinds of buffer overflow problems.
553 These should probably be used for compiling Perl whenever available,
554 Configure and/or hints files should be adjusted to probe for the
555 availability of these features and enable them as appropriate.
557 =head2 Arenas for GPs? For MAGIC?
559 C<struct gp> and C<struct magic> are both currently allocated by C<malloc>.
560 It might be a speed or memory saving to change to using arenas. Or it might
561 not. It would need some suitable benchmarking first. In particular, C<GP>s
562 can probably be changed with minimal compatibility impact (probably nothing
563 outside of the core, or even outside of F<gv.c> allocates them), but they
564 probably aren't allocated/deallocated often enough for a speed saving. Whereas
565 C<MAGIC> is allocated/deallocated more often, but in turn, is also something
566 more externally visible, so changing the rules here may bite external code.
570 Several SV body structs are now the same size, notably PVMG and PVGV, PVAV and
571 PVHV, and PVCV and PVFM. It should be possible to allocate and return same
572 sized bodies from the same actual arena, rather than maintaining one arena for
573 each. This could save 4-6K per thread, of memory no longer tied up in the
574 not-yet-allocated part of an arena.
577 =head1 Tasks that need a knowledge of XS
579 These tasks would need C knowledge, and roughly the level of knowledge of
580 the perl API that comes from writing modules that use XS to interface to
583 =head2 Write an XS cookbook
585 Create pod/perlxscookbook.pod with short, task-focused 'recipes' in XS that
586 demonstrate common tasks and good practices. (Some of these might be
587 extracted from perlguts.) The target audience should be XS novices, who need
588 more examples than perlguts but something less overwhelming than perlapi.
589 Recipes should provide "one pretty good way to do it" instead of TIMTOWTDI.
591 Rather than focusing on interfacing Perl to C libraries, such a cookbook
592 should probably focus on how to optimize Perl routines by re-writing them
593 in XS. This will likely be more motivating to those who mostly work in
594 Perl but are looking to take the next step into XS.
596 Deconstructing and explaining some simpler XS modules could be one way to
597 bootstrap a cookbook. (List::Util? Class::XSAccessor? Tree::Ternary_XS?)
598 Another option could be deconstructing the implementation of some simpler
601 =head2 Allow XSUBs to inline themselves as OPs
603 For a simple XSUB, often the subroutine dispatch takes more time than the
604 XSUB itself. The tokeniser already has the ability to inline constant
605 subroutines - it would be good to provide a way to inline other subroutines.
607 Specifically, simplest approach looks to be to allow an XSUB to provide an
608 alternative implementation of itself as a custom OP. A new flag bit in
609 C<CvFLAGS()> would signal to the peephole optimiser to take an optree
612 b <@> leave[1 ref] vKP/REFC ->(end)
614 2 <;> nextstate(main 1 -e:1) v:{ ->3
615 a <2> sassign vKS/2 ->b
616 8 <1> entersub[t2] sKS/TARG,1 ->9
619 4 <$> const(IV 1) sM ->5
620 6 <1> rv2av[t1] lKM/1 ->7
622 - <1> ex-rv2cv sK ->-
623 7 <$> gv(*x) s/EARLYCV ->8
624 - <1> ex-rv2sv sKRM*/1 ->a
627 perform the symbol table lookup of C<rv2cv> and C<gv(*x)>, locate the
628 pointer to the custom OP that provides the direct implementation, and re-
629 write the optree something like:
631 b <@> leave[1 ref] vKP/REFC ->(end)
633 2 <;> nextstate(main 1 -e:1) v:{ ->3
634 a <2> sassign vKS/2 ->b
638 4 <$> const(IV 1) sM ->5
639 6 <1> rv2av[t1] lKM/1 ->7
641 - <1> ex-rv2cv sK ->-
642 - <$> ex-gv(*x) s/EARLYCV ->7
643 - <1> ex-rv2sv sKRM*/1 ->a
646 I<i.e.> the C<gv(*)> OP has been nulled and spliced out of the execution
647 path, and the C<entersub> OP has been replaced by the custom op.
649 This approach should provide a measurable speed up to simple XSUBs inside
650 tight loops. Initially one would have to write the OP alternative
651 implementation by hand, but it's likely that this should be reasonably
652 straightforward for the type of XSUB that would benefit the most. Longer
653 term, once the run-time implementation is proven, it should be possible to
654 progressively update ExtUtils::ParseXS to generate OP implementations for
657 =head2 Remove the use of SVs as temporaries in dump.c
659 F<dump.c> contains debugging routines to dump out the contains of perl data
660 structures, such as C<SV>s, C<AV>s and C<HV>s. Currently, the dumping code
661 B<uses> C<SV>s for its temporary buffers, which was a logical initial
662 implementation choice, as they provide ready made memory handling.
664 However, they also lead to a lot of confusion when it happens that what you're
665 trying to debug is seen by the code in F<dump.c>, correctly or incorrectly, as
666 a temporary scalar it can use for a temporary buffer. It's also not possible
667 to dump scalars before the interpreter is properly set up, such as during
668 ithreads cloning. It would be good to progressively replace the use of scalars
669 as string accumulation buffers with something much simpler, directly allocated
670 by C<malloc>. The F<dump.c> code is (or should be) only producing 7 bit
671 US-ASCII, so output character sets are not an issue.
673 Producing and proving an internal simple buffer allocation would make it easier
674 to re-write the internals of the PerlIO subsystem to avoid using C<SV>s for
675 B<its> buffers, use of which can cause problems similar to those of F<dump.c>,
678 =head2 safely supporting POSIX SA_SIGINFO
680 Some years ago Jarkko supplied patches to provide support for the POSIX
681 SA_SIGINFO feature in Perl, passing the extra data to the Perl signal handler.
683 Unfortunately, it only works with "unsafe" signals, because under safe
684 signals, by the time Perl gets to run the signal handler, the extra
685 information has been lost. Moreover, it's not easy to store it somewhere,
686 as you can't call mutexs, or do anything else fancy, from inside a signal
689 So it strikes me that we could provide safe SA_SIGINFO support
695 Provide global variables for two file descriptors
699 When the first request is made via C<sigaction> for C<SA_SIGINFO>, create a
700 pipe, store the reader in one, the writer in the other
704 In the "safe" signal handler (C<Perl_csighandler()>/C<S_raise_signal()>), if
705 the C<siginfo_t> pointer non-C<NULL>, and the writer file handle is open,
711 serialise signal number, C<struct siginfo_t> (or at least the parts we care
712 about) into a small auto char buff
716 C<write()> that (non-blocking) to the writer fd
722 if it writes 100%, flag the signal in a counter of "signals on the pipe" akin
723 to the current per-signal-number counts
727 if it writes 0%, assume the pipe is full. Flag the data as lost?
731 if it writes partially, croak a panic, as your OS is broken.
739 in the regular C<PERL_ASYNC_CHECK()> processing, if there are "signals on
740 the pipe", read the data out, deserialise, build the Perl structures on
741 the stack (code in C<Perl_sighandler()>, the "unsafe" handler), and call as
746 I think that this gets us decent C<SA_SIGINFO> support, without the current risk
747 of running Perl code inside the signal handler context. (With all the dangers
748 of things like C<malloc> corruption that that currently offers us)
750 For more information see the thread starting with this message:
751 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-03/msg00305.html
753 =head2 autovivification
755 Make all autovivification consistent w.r.t LVALUE/RVALUE and strict/no strict;
757 This task is incremental - even a little bit of work on it will help.
759 =head2 Unicode in Filenames
761 chdir, chmod, chown, chroot, exec, glob, link, lstat, mkdir, open,
762 opendir, qx, readdir, readlink, rename, rmdir, stat, symlink, sysopen,
763 system, truncate, unlink, utime, -X. All these could potentially accept
764 Unicode filenames either as input or output (and in the case of system
765 and qx Unicode in general, as input or output to/from the shell).
766 Whether a filesystem - an operating system pair understands Unicode in
769 Known combinations that have some level of understanding include
770 Microsoft NTFS, Apple HFS+ (In Mac OS 9 and X) and Apple UFS (in Mac
771 OS X), NFS v4 is rumored to be Unicode, and of course Plan 9. How to
772 create Unicode filenames, what forms of Unicode are accepted and used
773 (UCS-2, UTF-16, UTF-8), what (if any) is the normalization form used,
774 and so on, varies. Finding the right level of interfacing to Perl
775 requires some thought. Remember that an OS does not implicate a
778 (The Windows -C command flag "wide API support" has been at least
779 temporarily retired in 5.8.1, and the -C has been repurposed, see
782 Most probably the right way to do this would be this:
783 L</"Virtualize operating system access">.
785 =head2 Unicode in %ENV
787 Currently the %ENV entries are always byte strings.
788 See L</"Virtualize operating system access">.
790 =head2 Unicode and glob()
792 Currently glob patterns and filenames returned from File::Glob::glob()
793 are always byte strings. See L</"Virtualize operating system access">.
795 =head2 use less 'memory'
797 Investigate trade offs to switch out perl's choices on memory usage.
798 Particularly perl should be able to give memory back.
800 This task is incremental - even a little bit of work on it will help.
802 =head2 Re-implement C<:unique> in a way that is actually thread-safe
804 The old implementation made bad assumptions on several levels. A good 90%
805 solution might be just to make C<:unique> work to share the string buffer
806 of SvPVs. That way large constant strings can be shared between ithreads,
807 such as the configuration information in F<Config>.
809 =head2 Make tainting consistent
811 Tainting would be easier to use if it didn't take documented shortcuts and
812 allow taint to "leak" everywhere within an expression.
814 =head2 readpipe(LIST)
816 system() accepts a LIST syntax (and a PROGRAM LIST syntax) to avoid
817 running a shell. readpipe() (the function behind qx//) could be similarly
820 =head2 Audit the code for destruction ordering assumptions
824 /* Need to check SvMAGICAL, as during global destruction it may be that
825 AvARYLEN(av) has been freed before av, and hence the SvANY() pointer
826 is now part of the linked list of SV heads, rather than pointing to
827 the original body. */
828 /* FIXME - audit the code for other bugs like this one. */
830 adding the C<SvMAGICAL> check to
832 if (AvARYLEN(av) && SvMAGICAL(AvARYLEN(av))) {
833 MAGIC *mg = mg_find (AvARYLEN(av), PERL_MAGIC_arylen);
835 Go through the core and look for similar assumptions that SVs have particular
836 types, as all bets are off during global destruction.
838 =head2 Extend PerlIO and PerlIO::Scalar
840 PerlIO::Scalar doesn't know how to truncate(). Implementing this
841 would require extending the PerlIO vtable.
843 Similarly the PerlIO vtable doesn't know about formats (write()), or
844 about stat(), or chmod()/chown(), utime(), or flock().
846 (For PerlIO::Scalar it's hard to see what e.g. mode bits or ownership
849 PerlIO doesn't do directories or symlinks, either: mkdir(), rmdir(),
850 opendir(), closedir(), seekdir(), rewinddir(), glob(); symlink(),
853 See also L</"Virtualize operating system access">.
855 =head2 -C on the #! line
857 It should be possible to make -C work correctly if found on the #! line,
858 given that all perl command line options are strict ASCII, and -C changes
859 only the interpretation of non-ASCII characters, and not for the script file
860 handle. To make it work needs some investigation of the ordering of function
861 calls during startup, and (by implication) a bit of tweaking of that order.
863 =head2 Organize error messages
865 Perl's diagnostics (error messages, see L<perldiag>) could use
866 reorganizing and formalizing so that each error message has its
867 stable-for-all-eternity unique id, categorized by severity, type, and
868 subsystem. (The error messages would be listed in a datafile outside
869 of the Perl source code, and the source code would only refer to the
870 messages by the id.) This clean-up and regularizing should apply
871 for all croak() messages.
873 This would enable all sorts of things: easier translation/localization
874 of the messages (though please do keep in mind the caveats of
875 L<Locale::Maketext> about too straightforward approaches to
876 translation), filtering by severity, and instead of grepping for a
877 particular error message one could look for a stable error id. (Of
878 course, changing the error messages by default would break all the
879 existing software depending on some particular error message...)
881 This kind of functionality is known as I<message catalogs>. Look for
882 inspiration for example in the catgets() system, possibly even use it
883 if available-- but B<only> if available, all platforms will B<not>
886 For the really pure at heart, consider extending this item to cover
887 also the warning messages (see L<perllexwarn>, C<warnings.pl>).
889 =head1 Tasks that need a knowledge of the interpreter
891 These tasks would need C knowledge, and knowledge of how the interpreter works,
892 or a willingness to learn.
894 =head2 forbid labels with keyword names
896 Currently C<goto keyword> "computes" the label value:
898 $ perl -e 'goto print'
899 Can't find label 1 at -e line 1.
901 It is controversial if the right way to avoid the confusion is to forbid
902 labels with keyword names, or if it would be better to always treat
903 bareword expressions after a "goto" as a label and never as a keyword.
905 =head2 truncate() prototype
907 The prototype of truncate() is currently C<$$>. It should probably
908 be C<*$> instead. (This is changed in F<opcode.pl>)
910 =head2 decapsulation of smart match argument
912 Currently C<$foo ~~ $object> will die with the message "Smart matching a
913 non-overloaded object breaks encapsulation". It would be nice to allow
914 to bypass this by using explicitly the syntax C<$foo ~~ %$object> or
917 =head2 error reporting of [$a ; $b]
919 Using C<;> inside brackets is a syntax error, and we don't propose to change
920 that by giving it any meaning. However, it's not reported very helpfully:
922 $ perl -e '$a = [$b; $c];'
923 syntax error at -e line 1, near "$b;"
924 syntax error at -e line 1, near "$c]"
925 Execution of -e aborted due to compilation errors.
927 It should be possible to hook into the tokeniser or the lexer, so that when a
928 C<;> is parsed where it is not legal as a statement terminator (ie inside
929 C<{}> used as a hashref, C<[]> or C<()>) it issues an error something like
930 I<';' isn't legal inside an expression - if you need multiple statements use a
931 do {...} block>. See the thread starting at
932 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-09/msg00573.html
934 =head2 lexicals used only once
938 $ perl -we '$pie = 42'
939 Name "main::pie" used only once: possible typo at -e line 1.
943 $ perl -we 'my $pie = 42'
945 Logically all lexicals used only once should warn, if the user asks for
946 warnings. An unworked RT ticket (#5087) has been open for almost seven
947 years for this discrepancy.
951 The handling of Unicode is unclean in many places. In the regex engine
952 there are especially many problems. The swash data structure could be
953 replaced my something better. Inversion lists and maps are likely
954 candidates. The whole Unicode database could be placed in-core for a
955 huge speed-up. Only minimal work was done on the optimizer when utf8
956 was added, with the result that the synthetic start class often will
957 fail to narrow down the possible choices when given non-Latin1 input.
959 =head2 Properly Unicode safe tokeniser and pads.
961 The tokeniser isn't actually very UTF-8 clean. C<use utf8;> is a hack -
962 variable names are stored in stashes as raw bytes, without the utf-8 flag
963 set. The pad API only takes a C<char *> pointer, so that's all bytes too. The
964 tokeniser ignores the UTF-8-ness of C<PL_rsfp>, or any SVs returned from
965 source filters. All this could be fixed.
967 =head2 state variable initialization in list context
969 Currently this is illegal:
971 state ($a, $b) = foo();
973 In Perl 6, C<state ($a) = foo();> and C<(state $a) = foo();> have different
974 semantics, which is tricky to implement in Perl 5 as currently they produce
975 the same opcode trees. The Perl 6 design is firm, so it would be good to
976 implement the necessary code in Perl 5. There are comments in
977 C<Perl_newASSIGNOP()> that show the code paths taken by various assignment
978 constructions involving state variables.
980 =head2 Implement $value ~~ 0 .. $range
982 It would be nice to extend the syntax of the C<~~> operator to also
983 understand numeric (and maybe alphanumeric) ranges.
985 =head2 A does() built-in
987 Like ref(), only useful. It would call the C<DOES> method on objects; it
988 would also tell whether something can be dereferenced as an
989 array/hash/etc., or used as a regexp, etc.
990 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-03/msg00481.html>
992 =head2 Tied filehandles and write() don't mix
994 There is no method on tied filehandles to allow them to be called back by
997 =head2 Propagate compilation hints to the debugger
999 Currently a debugger started with -dE on the command-line doesn't see the
1000 features enabled by -E. More generally hints (C<$^H> and C<%^H>) aren't
1001 propagated to the debugger. Probably it would be a good thing to propagate
1002 hints from the innermost non-C<DB::> scope: this would make code eval'ed
1003 in the debugger see the features (and strictures, etc.) currently in
1006 =head2 Attach/detach debugger from running program
1008 The old perltodo notes "With C<gdb>, you can attach the debugger to a running
1009 program if you pass the process ID. It would be good to do this with the Perl
1010 debugger on a running Perl program, although I'm not sure how it would be
1011 done." ssh and screen do this with named pipes in /tmp. Maybe we can too.
1013 =head2 LVALUE functions for lists
1015 The old perltodo notes that lvalue functions don't work for list or hash
1016 slices. This would be good to fix.
1018 =head2 regexp optimiser optional
1020 The regexp optimiser is not optional. It should configurable to be, to allow
1021 its performance to be measured, and its bugs to be easily demonstrated.
1023 =head2 C</w> regex modifier
1025 That flag would enable to match whole words, and also to interpolate
1026 arrays as alternations. With it, C</P/w> would be roughly equivalent to:
1028 do { local $"='|'; /\b(?:P)\b/ }
1030 See L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-01/msg00400.html>
1033 =head2 optional optimizer
1035 Make the peephole optimizer optional. Currently it performs two tasks as
1036 it walks the optree - genuine peephole optimisations, and necessary fixups of
1037 ops. It would be good to find an efficient way to switch out the
1038 optimisations whilst keeping the fixups.
1040 =head2 You WANT *how* many
1042 Currently contexts are void, scalar and list. split has a special mechanism in
1043 place to pass in the number of return values wanted. It would be useful to
1044 have a general mechanism for this, backwards compatible and little speed hit.
1045 This would allow proposals such as short circuiting sort to be implemented
1046 as a module on CPAN.
1048 =head2 lexical aliases
1050 Allow lexical aliases (maybe via the syntax C<my \$alias = \$foo>.
1052 =head2 entersub XS vs Perl
1054 At the moment pp_entersub is huge, and has code to deal with entering both
1055 perl and XS subroutines. Subroutine implementations rarely change between
1056 perl and XS at run time, so investigate using 2 ops to enter subs (one for
1057 XS, one for perl) and swap between if a sub is redefined.
1061 Self-ties are currently illegal because they caused too many segfaults. Maybe
1062 the causes of these could be tracked down and self-ties on all types
1065 =head2 Optimize away @_
1067 The old perltodo notes "Look at the "reification" code in C<av.c>".
1069 =head2 Virtualize operating system access
1071 Implement a set of "vtables" that virtualizes operating system access
1072 (open(), mkdir(), unlink(), readdir(), getenv(), etc.) At the very
1073 least these interfaces should take SVs as "name" arguments instead of
1074 bare char pointers; probably the most flexible and extensible way
1075 would be for the Perl-facing interfaces to accept HVs. The system
1076 needs to be per-operating-system and per-file-system
1077 hookable/filterable, preferably both from XS and Perl level
1078 (L<perlport/"Files and Filesystems"> is good reading at this point,
1079 in fact, all of L<perlport> is.)
1081 This has actually already been implemented (but only for Win32),
1082 take a look at F<iperlsys.h> and F<win32/perlhost.h>. While all Win32
1083 variants go through a set of "vtables" for operating system access,
1084 non-Win32 systems currently go straight for the POSIX/Unix-style
1085 system/library call. Similar system as for Win32 should be
1086 implemented for all platforms. The existing Win32 implementation
1087 probably does not need to survive alongside this proposed new
1088 implementation, the approaches could be merged.
1090 What would this give us? One often-asked-for feature this would
1091 enable is using Unicode for filenames, and other "names" like %ENV,
1092 usernames, hostnames, and so forth.
1093 (See L<perlunicode/"When Unicode Does Not Happen">.)
1095 But this kind of virtualization would also allow for things like
1096 virtual filesystems, virtual networks, and "sandboxes" (though as long
1097 as dynamic loading of random object code is allowed, not very safe
1098 sandboxes since external code of course know not of Perl's vtables).
1099 An example of a smaller "sandbox" is that this feature can be used to
1100 implement per-thread working directories: Win32 already does this.
1102 See also L</"Extend PerlIO and PerlIO::Scalar">.
1104 =head2 Investigate PADTMP hash pessimisation
1106 The peephole optimiser converts constants used for hash key lookups to shared
1107 hash key scalars. Under ithreads, something is undoing this work.
1108 See http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-09/msg00793.html
1110 =head2 Store the current pad in the OP slab allocator
1113 I hope that I got that "current pad" part correct
1115 Currently we leak ops in various cases of parse failure. I suggested that we
1116 could solve this by always using the op slab allocator, and walking it to
1117 free ops. Dave comments that as some ops are already freed during optree
1118 creation one would have to mark which ops are freed, and not double free them
1119 when walking the slab. He notes that one problem with this is that for some ops
1120 you have to know which pad was current at the time of allocation, which does
1121 change. I suggested storing a pointer to the current pad in the memory allocated
1122 for the slab, and swapping to a new slab each time the pad changes. Dave thinks
1123 that this would work.
1125 =head2 repack the optree
1127 Repacking the optree after execution order is determined could allow
1128 removal of NULL ops, and optimal ordering of OPs with respect to cache-line
1129 filling. The slab allocator could be reused for this purpose. I think that
1130 the best way to do this is to make it an optional step just before the
1131 completed optree is attached to anything else, and to use the slab allocator
1132 unchanged, so that freeing ops is identical whether or not this step runs.
1133 Note that the slab allocator allocates ops downwards in memory, so one would
1134 have to actually "allocate" the ops in reverse-execution order to get them
1135 contiguous in memory in execution order.
1137 See http://www.nntp.perl.org/group/perl.perl5.porters/2007/12/msg131975.html
1139 Note that running this copy, and then freeing all the old location ops would
1140 cause their slabs to be freed, which would eliminate possible memory wastage if
1141 the previous suggestion is implemented, and we swap slabs more frequently.
1143 =head2 eliminate incorrect line numbers in warnings
1151 } elsif ($undef == 0) {
1154 used to produce this output:
1156 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1157 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1159 where the line of the second warning was misreported - it should be line 5.
1160 Rafael fixed this - the problem arose because there was no nextstate OP
1161 between the execution of the C<if> and the C<elsif>, hence C<PL_curcop> still
1162 reports that the currently executing line is line 4. The solution was to inject
1163 a nextstate OPs for each C<elsif>, although it turned out that the nextstate
1164 OP needed to be a nulled OP, rather than a live nextstate OP, else other line
1165 numbers became misreported. (Jenga!)
1167 The problem is more general than C<elsif> (although the C<elsif> case is the
1168 most common and the most confusing). Ideally this code
1178 would produce this output
1180 Use of uninitialized value $undef in addition (+) at wrong.pl line 4.
1181 Use of uninitialized value $undef in addition (+) at wrong.pl line 7.
1183 (rather than lines 4 and 5), but this would seem to require every OP to carry
1184 (at least) line number information.
1186 What might work is to have an optional line number in memory just before the
1187 BASEOP structure, with a flag bit in the op to say whether it's present.
1188 Initially during compile every OP would carry its line number. Then add a late
1189 pass to the optimiser (potentially combined with L</repack the optree>) which
1190 looks at the two ops on every edge of the graph of the execution path. If
1191 the line number changes, flags the destination OP with this information.
1192 Once all paths are traced, replace every op with the flag with a
1193 nextstate-light op (that just updates C<PL_curcop>), which in turn then passes
1194 control on to the true op. All ops would then be replaced by variants that
1195 do not store the line number. (Which, logically, why it would work best in
1196 conjunction with L</repack the optree>, as that is already copying/reallocating
1199 (Although I should note that we're not certain that doing this for the general
1202 =head2 optimize tail-calls
1204 Tail-calls present an opportunity for broadly applicable optimization;
1205 anywhere that C<< return foo(...) >> is called, the outer return can
1206 be replaced by a goto, and foo will return directly to the outer
1207 caller, saving (conservatively) 25% of perl's call&return cost, which
1208 is relatively higher than in C. The scheme language is known to do
1209 this heavily. B::Concise provides good insight into where this
1210 optimization is possible, ie anywhere entersub,leavesub op-sequence
1213 perl -MO=Concise,-exec,a,b,-main -e 'sub a{ 1 }; sub b {a()}; b(2)'
1215 Bottom line on this is probably a new pp_tailcall function which
1216 combines the code in pp_entersub, pp_leavesub. This should probably
1217 be done 1st in XS, and using B::Generate to patch the new OP into the
1220 =head2 Add C<00dddd>
1222 It has been proposed that octal constants be specifiable through the syntax
1223 C<0oddddd>, parallel to the existing construct to specify hex constants
1228 Tasks that will get your name mentioned in the description of the "Highlights
1231 =head2 make ithreads more robust
1233 Generally make ithreads more robust. See also L</iCOW>
1235 This task is incremental - even a little bit of work on it will help, and
1236 will be greatly appreciated.
1238 One bit would be to determine how to clone directory handles on systems
1239 without a C<fchdir> function (in sv.c:Perl_dirp_dup).
1241 Fix Perl_sv_dup, et al so that threads can return objects.
1245 Sarathy and Arthur have a proposal for an improved Copy On Write which
1246 specifically will be able to COW new ithreads. If this can be implemented
1247 it would be a good thing.
1249 =head2 (?{...}) closures in regexps
1251 Fix (or rewrite) the implementation of the C</(?{...})/> closures.
1253 =head2 Add class set operations to regexp engine
1255 Apparently these are quite useful. Anyway, Jeffery Friedl wants them.
1257 demerphq has this on his todo list, but right at the bottom.
1260 =head1 Tasks for microperl
1263 [ Each and every one of these may be obsolete, but they were listed
1264 in the old Todo.micro file]
1267 =head2 make creating uconfig.sh automatic
1269 =head2 make creating Makefile.micro automatic
1271 =head2 do away with fork/exec/wait?
1273 (system, popen should be enough?)
1275 =head2 some of the uconfig.sh really needs to be probed (using cc) in buildtime:
1277 (uConfigure? :-) native datatype widths and endianness come to mind