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 Test that regen.pl was run
44 There are various generated files shipped with the perl distribution, for
45 things like header files generate from data. The generation scripts are
46 written in perl, and all can be run by F<regen.pl>. However, because they're
47 written in perl, we can't run them before we've built perl. We can't run them
48 as part of the F<Makefile>, because changing files underneath F<make> confuses
49 it completely, and we don't want to run them automatically anyway, as they
50 change files shipped by the distribution, something we seek not do to.
52 If someone changes the data, but forgets to re-run F<regen.pl> then the
53 generated files are out of sync. It would be good to have a test in
54 F<t/porting> that checks that the generated files are in sync, and fails
55 otherwise, to alert someone before they make a poor commit. I suspect that this
56 would require adapting the scripts run from F<regen.pl> to have dry-run
57 options, and invoking them with these, or by refactoring them into a library
58 that does the generation, which can be called by the scripts, and by the test.
60 =head2 Automate perldelta generation
62 The perldelta file accompanying each release summaries the major changes.
63 It's mostly manually generated currently, but some of that could be
64 automated with a bit of perl, specifically the generation of
68 =item Modules and Pragmata
70 =item New Documentation
76 See F<Porting/how_to_write_a_perldelta.pod> for details.
78 =head2 Remove duplication of test setup.
80 Schwern notes, that there's duplication of code - lots and lots of tests have
81 some variation on the big block of C<$Is_Foo> checks. We can safely put this
82 into a file, change it to build an C<%Is> hash and require it. Maybe just put
83 it into F<test.pl>. Throw in the handy tainting subroutines.
85 =head2 POD -E<gt> HTML conversion in the core still sucks
87 Which is crazy given just how simple POD purports to be, and how simple HTML
88 can be. It's not actually I<as> simple as it sounds, particularly with the
89 flexibility POD allows for C<=item>, but it would be good to improve the
90 visual appeal of the HTML generated, and to avoid it having any validation
91 errors. See also L</make HTML install work>, as the layout of installation tree
92 is needed to improve the cross-linking.
94 The addition of C<Pod::Simple> and its related modules may make this task
97 =head2 Make ExtUtils::ParseXS use strict;
99 F<lib/ExtUtils/ParseXS.pm> contains this line
101 # use strict; # One of these days...
103 Simply uncomment it, and fix all the resulting issues :-)
105 The more practical approach, to break the task down into manageable chunks, is
106 to work your way though the code from bottom to top, or if necessary adding
107 extra C<{ ... }> blocks, and turning on strict within them.
109 =head2 Make Schwern poorer
111 We should have tests for everything. When all the core's modules are tested,
112 Schwern has promised to donate to $500 to TPF. We may need volunteers to
113 hold him upside down and shake vigorously in order to actually extract the
116 =head2 Improve the coverage of the core tests
118 Use Devel::Cover to ascertain the core modules' test coverage, then add
119 tests that are currently missing.
123 A full test suite for the B module would be nice.
125 =head2 A decent benchmark
127 C<perlbench> seems impervious to any recent changes made to the perl core. It
128 would be useful to have a reasonable general benchmarking suite that roughly
129 represented what current perl programs do, and measurably reported whether
130 tweaks to the core improve, degrade or don't really affect performance, to
131 guide people attempting to optimise the guts of perl. Gisle would welcome
132 new tests for perlbench.
134 =head2 fix tainting bugs
136 Fix the bugs revealed by running the test suite with the C<-t> switch (via
137 C<make test.taintwarn>).
139 =head2 Dual life everything
141 As part of the "dists" plan, anything that doesn't belong in the smallest perl
142 distribution needs to be dual lifed. Anything else can be too. Figure out what
143 changes would be needed to package that module and its tests up for CPAN, and
144 do so. Test it with older perl releases, and fix the problems you find.
146 To make a minimal perl distribution, it's useful to look at
147 F<t/lib/commonsense.t>.
149 =head2 POSIX memory footprint
151 Ilya observed that use POSIX; eats memory like there's no tomorrow, and at
152 various times worked to cut it down. There is probably still fat to cut out -
153 for example POSIX passes Exporter some very memory hungry data structures.
155 =head2 embed.pl/makedef.pl
157 There is a script F<embed.pl> that generates several header files to prefix
158 all of Perl's symbols in a consistent way, to provide some semblance of
159 namespace support in C<C>. Functions are declared in F<embed.fnc>, variables
160 in F<interpvar.h>. Quite a few of the functions and variables
161 are conditionally declared there, using C<#ifdef>. However, F<embed.pl>
162 doesn't understand the C macros, so the rules about which symbols are present
163 when is duplicated in F<makedef.pl>. Writing things twice is bad, m'kay.
164 It would be good to teach C<embed.pl> to understand the conditional
165 compilation, and hence remove the duplication, and the mistakes it has caused.
167 =head2 use strict; and AutoLoad
169 Currently if you write
172 use AutoLoader 'AUTOLOAD';
177 print join (' ', No, strict, here), "!\n";
180 then C<use strict;> isn't in force within the autoloaded subroutines. It would
181 be more consistent (and less surprising) to arrange for all lexical pragmas
182 in force at the __END__ block to be in force within each autoloaded subroutine.
184 There's a similar problem with SelfLoader.
186 =head2 profile installman
188 The F<installman> script is slow. All it is doing text processing, which we're
189 told is something Perl is good at. So it would be nice to know what it is doing
190 that is taking so much CPU, and where possible address it.
192 =head2 enable lexical enabling/disabling of inidvidual warnings
194 Currently, warnings can only be enabled or disabled by category. There
195 are times when it would be useful to quash a single warning, not a
198 =head1 Tasks that need a little sysadmin-type knowledge
200 Or if you prefer, tasks that you would learn from, and broaden your skills
203 =head2 make HTML install work
205 There is an C<installhtml> target in the Makefile. It's marked as
206 "experimental". It would be good to get this tested, make it work reliably, and
207 remove the "experimental" tag. This would include
213 Checking that cross linking between various parts of the documentation works.
214 In particular that links work between the modules (files with POD in F<lib/>)
215 and the core documentation (files in F<pod/>)
219 Work out how to split C<perlfunc> into chunks, preferably one per function
220 group, preferably with general case code that could be used elsewhere.
221 Challenges here are correctly identifying the groups of functions that go
222 together, and making the right named external cross-links point to the right
223 page. Things to be aware of are C<-X>, groups such as C<getpwnam> to
224 C<endservent>, two or more C<=items> giving the different parameter lists, such
227 =item substr EXPR,OFFSET,LENGTH,REPLACEMENT
228 =item substr EXPR,OFFSET,LENGTH
229 =item substr EXPR,OFFSET
231 and different parameter lists having different meanings. (eg C<select>)
235 =head2 compressed man pages
237 Be able to install them. This would probably need a configure test to see how
238 the system does compressed man pages (same directory/different directory?
239 same filename/different filename), as well as tweaking the F<installman> script
240 to compress as necessary.
242 =head2 Add a code coverage target to the Makefile
244 Make it easy for anyone to run Devel::Cover on the core's tests. The steps
245 to do this manually are roughly
251 do a normal C<Configure>, but include Devel::Cover as a module to install
252 (see L<INSTALL> for how to do this)
260 cd t; HARNESS_PERL_SWITCHES=-MDevel::Cover ./perl -I../lib harness
264 Process the resulting Devel::Cover database
268 This just give you the coverage of the F<.pm>s. To also get the C level
275 Additionally tell C<Configure> to use the appropriate C compiler flags for
282 (instead of C<make perl>)
286 After running the tests run C<gcov> to generate all the F<.gcov> files.
287 (Including down in the subdirectories of F<ext/>
291 (From the top level perl directory) run C<gcov2perl> on all the C<.gcov> files
292 to get their stats into the cover_db directory.
296 Then process the Devel::Cover database
300 It would be good to add a single switch to C<Configure> to specify that you
301 wanted to perform perl level coverage, and another to specify C level
302 coverage, and have C<Configure> and the F<Makefile> do all the right things
305 =head2 Make Config.pm cope with differences between built and installed perl
307 Quite often vendors ship a perl binary compiled with their (pay-for)
308 compilers. People install a free compiler, such as gcc. To work out how to
309 build extensions, Perl interrogates C<%Config>, so in this situation
310 C<%Config> describes compilers that aren't there, and extension building
311 fails. This forces people into choosing between re-compiling perl themselves
312 using the compiler they have, or only using modules that the vendor ships.
314 It would be good to find a way teach C<Config.pm> about the installation setup,
315 possibly involving probing at install time or later, so that the C<%Config> in
316 a binary distribution better describes the installed machine, when the
317 installed machine differs from the build machine in some significant way.
319 =head2 linker specification files
321 Some platforms mandate that you provide a list of a shared library's external
322 symbols to the linker, so the core already has the infrastructure in place to
323 do this for generating shared perl libraries. My understanding is that the
324 GNU toolchain can accept an optional linker specification file, and restrict
325 visibility just to symbols declared in that file. It would be good to extend
326 F<makedef.pl> to support this format, and to provide a means within
327 C<Configure> to enable it. This would allow Unix users to test that the
328 export list is correct, and to build a perl that does not pollute the global
329 namespace with private symbols, and will fail in the same way as msvc or mingw
330 builds or when using PERL_DL_NONLAZY=1.
332 =head2 Cross-compile support
334 Currently C<Configure> understands C<-Dusecrosscompile> option. This option
335 arranges for building C<miniperl> for TARGET machine, so this C<miniperl> is
336 assumed then to be copied to TARGET machine and used as a replacement of full
339 This could be done little differently. Namely C<miniperl> should be built for
340 HOST and then full C<perl> with extensions should be compiled for TARGET.
341 This, however, might require extra trickery for %Config: we have one config
342 first for HOST and then another for TARGET. Tools like MakeMaker will be
343 mightily confused. Having around two different types of executables and
344 libraries (HOST and TARGET) makes life interesting for Makefiles and
345 shell (and Perl) scripts. There is $Config{run}, normally empty, which
346 can be used as an execution wrapper. Also note that in some
347 cross-compilation/execution environments the HOST and the TARGET do
348 not see the same filesystem(s), the $Config{run} may need to do some
349 file/directory copying back and forth.
353 Make F<pod/roffitall> be updated by F<pod/buildtoc>.
355 =head2 Split "linker" from "compiler"
357 Right now, Configure probes for two commands, and sets two variables:
361 =item * C<cc> (in F<cc.U>)
363 This variable holds the name of a command to execute a C compiler which
364 can resolve multiple global references that happen to have the same
365 name. Usual values are F<cc> and F<gcc>.
366 Fervent ANSI compilers may be called F<c89>. AIX has F<xlc>.
368 =item * C<ld> (in F<dlsrc.U>)
370 This variable indicates the program to be used to link
371 libraries for dynamic loading. On some systems, it is F<ld>.
372 On ELF systems, it should be C<$cc>. Mostly, we'll try to respect
373 the hint file setting.
377 There is an implicit historical assumption from around Perl5.000alpha
378 something, that C<$cc> is also the correct command for linking object files
379 together to make an executable. This may be true on Unix, but it's not true
380 on other platforms, and there are a maze of work arounds in other places (such
381 as F<Makefile.SH>) to cope with this.
383 Ideally, we should create a new variable to hold the name of the executable
384 linker program, probe for it in F<Configure>, and centralise all the special
385 case logic there or in hints files.
387 A small bikeshed issue remains - what to call it, given that C<$ld> is already
388 taken (arguably for the wrong thing now, but on SunOS 4.1 it is the command
389 for creating dynamically-loadable modules) and C<$link> could be confused with
390 the Unix command line executable of the same name, which does something
391 completely different. Andy Dougherty makes the counter argument "In parrot, I
392 tried to call the command used to link object files and libraries into an
393 executable F<link>, since that's what my vaguely-remembered DOS and VMS
394 experience suggested. I don't think any real confusion has ensued, so it's
395 probably a reasonable name for perl5 to use."
397 "Alas, I've always worried that introducing it would make things worse,
398 since now the module building utilities would have to look for
399 C<$Config{link}> and institute a fall-back plan if it weren't found."
400 Although I can see that as confusing, given that C<$Config{d_link}> is true
401 when (hard) links are available.
403 =head2 Configure Windows using PowerShell
405 Currently, Windows uses hard-coded config files based to build the
406 config.h for compiling Perl. Makefiles are also hard-coded and need to be
407 hand edited prior to building Perl. While this makes it easy to create a perl.exe
408 that works across multiple Windows versions, being able to accurately
409 configure a perl.exe for a specific Windows versions and VS C++ would be
410 a nice enhancement. With PowerShell available on Windows XP and up, this
411 may now be possible. Step 1 might be to investigate whether this is possible
412 and use this to clean up our current makefile situation. Step 2 would be to
413 see if there would be a way to use our existing metaconfig units to configure a
414 Windows Perl or whether we go in a separate direction and make it so. Of
415 course, we all know what step 3 is.
417 =head2 decouple -g and -DDEBUGGING
419 Currently F<Configure> automatically adds C<-DDEBUGGING> to the C compiler
420 flags if it spots C<-g> in the optimiser flags. The pre-processor directive
421 C<DEBUGGING> enables F<perl>'s command line C<-D> options, but in the process
422 makes F<perl> slower. It would be good to disentangle this logic, so that
423 C-level debugging with C<-g> and Perl level debugging with C<-D> can easily
424 be enabled independently.
426 =head1 Tasks that need a little C knowledge
428 These tasks would need a little C knowledge, but don't need any specific
429 background or experience with XS, or how the Perl interpreter works
431 =head2 Weed out needless PERL_UNUSED_ARG
433 The C code uses the macro C<PERL_UNUSED_ARG> to stop compilers warning about
434 unused arguments. Often the arguments can't be removed, as there is an
435 external constraint that determines the prototype of the function, so this
436 approach is valid. However, there are some cases where C<PERL_UNUSED_ARG>
437 could be removed. Specifically
443 The prototypes of (nearly all) static functions can be changed
447 Unused arguments generated by short cut macros are wasteful - the short cut
448 macro used can be changed.
452 =head2 Modernize the order of directories in @INC
454 The way @INC is laid out by default, one cannot upgrade core (dual-life)
455 modules without overwriting files. This causes problems for binary
456 package builders. One possible proposal is laid out in this
458 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2002-04/msg02380.html>.
462 Natively 64-bit systems need neither -Duse64bitint nor -Duse64bitall.
463 On these systems, it might be the default compilation mode, and there
464 is currently no guarantee that passing no use64bitall option to the
465 Configure process will build a 32bit perl. Implementing -Duse32bit*
466 options would be nice for perl 5.12.
468 =head2 Profile Perl - am I hot or not?
470 The Perl source code is stable enough that it makes sense to profile it,
471 identify and optimise the hotspots. It would be good to measure the
472 performance of the Perl interpreter using free tools such as cachegrind,
473 gprof, and dtrace, and work to reduce the bottlenecks they reveal.
475 As part of this, the idea of F<pp_hot.c> is that it contains the I<hot> ops,
476 the ops that are most commonly used. The idea is that by grouping them, their
477 object code will be adjacent in the executable, so they have a greater chance
478 of already being in the CPU cache (or swapped in) due to being near another op
481 Except that it's not clear if these really are the most commonly used ops. So
482 as part of exercising your skills with coverage and profiling tools you might
483 want to determine what ops I<really> are the most commonly used. And in turn
484 suggest evictions and promotions to achieve a better F<pp_hot.c>.
486 One piece of Perl code that might make a good testbed is F<installman>.
488 =head2 Allocate OPs from arenas
490 Currently all new OP structures are individually malloc()ed and free()d.
491 All C<malloc> implementations have space overheads, and are now as fast as
492 custom allocates so it would both use less memory and less CPU to allocate
493 the various OP structures from arenas. The SV arena code can probably be
496 Note that Configuring perl with C<-Accflags=-DPL_OP_SLAB_ALLOC> will use
497 Perl_Slab_alloc() to pack optrees into a contiguous block, which is
498 probably superior to the use of OP arenas, esp. from a cache locality
499 standpoint. See L<Profile Perl - am I hot or not?>.
501 =head2 Improve win32/wince.c
503 Currently, numerous functions look virtually, if not completely,
504 identical in both C<win32/wince.c> and C<win32/win32.c> files, which can't
507 =head2 Use secure CRT functions when building with VC8 on Win32
509 Visual C++ 2005 (VC++ 8.x) deprecated a number of CRT functions on the basis
510 that they were "unsafe" and introduced differently named secure versions of
511 them as replacements, e.g. instead of writing
513 FILE* f = fopen(__FILE__, "r");
518 errno_t err = fopen_s(&f, __FILE__, "r");
520 Currently, the warnings about these deprecations have been disabled by adding
521 -D_CRT_SECURE_NO_DEPRECATE to the CFLAGS. It would be nice to remove that
522 warning suppressant and actually make use of the new secure CRT functions.
524 There is also a similar issue with POSIX CRT function names like fileno having
525 been deprecated in favour of ISO C++ conformant names like _fileno. These
526 warnings are also currently suppressed by adding -D_CRT_NONSTDC_NO_DEPRECATE. It
527 might be nice to do as Microsoft suggest here too, although, unlike the secure
528 functions issue, there is presumably little or no benefit in this case.
530 =head2 Fix POSIX::access() and chdir() on Win32
532 These functions currently take no account of DACLs and therefore do not behave
533 correctly in situations where access is restricted by DACLs (as opposed to the
534 read-only attribute).
536 Furthermore, POSIX::access() behaves differently for directories having the
537 read-only attribute set depending on what CRT library is being used. For
538 example, the _access() function in the VC6 and VC7 CRTs (wrongly) claim that
539 such directories are not writable, whereas in fact all directories are writable
540 unless access is denied by DACLs. (In the case of directories, the read-only
541 attribute actually only means that the directory cannot be deleted.) This CRT
542 bug is fixed in the VC8 and VC9 CRTs (but, of course, the directory may still
543 not actually be writable if access is indeed denied by DACLs).
545 For the chdir() issue, see ActiveState bug #74552:
546 http://bugs.activestate.com/show_bug.cgi?id=74552
548 Therefore, DACLs should be checked both for consistency across CRTs and for
551 (Note that perl's -w operator should not be modified to check DACLs. It has
552 been written so that it reflects the state of the read-only attribute, even
553 for directories (whatever CRT is being used), for symmetry with chmod().)
555 =head2 strcat(), strcpy(), strncat(), strncpy(), sprintf(), vsprintf()
557 Maybe create a utility that checks after each libperl.a creation that
558 none of the above (nor sprintf(), vsprintf(), or *SHUDDER* gets())
559 ever creep back to libperl.a.
561 nm libperl.a | ./miniperl -alne '$o = $F[0] if /:$/; print "$o $F[1]" if $F[0] eq "U" && $F[1] =~ /^(?:strn?c(?:at|py)|v?sprintf|gets)$/'
563 Note, of course, that this will only tell whether B<your> platform
564 is using those naughty interfaces.
566 =head2 -D_FORTIFY_SOURCE=2, -fstack-protector
568 Recent glibcs support C<-D_FORTIFY_SOURCE=2> and recent gcc
569 (4.1 onwards?) supports C<-fstack-protector>, both of which give
570 protection against various kinds of buffer overflow problems.
571 These should probably be used for compiling Perl whenever available,
572 Configure and/or hints files should be adjusted to probe for the
573 availability of these features and enable them as appropriate.
575 =head2 Arenas for GPs? For MAGIC?
577 C<struct gp> and C<struct magic> are both currently allocated by C<malloc>.
578 It might be a speed or memory saving to change to using arenas. Or it might
579 not. It would need some suitable benchmarking first. In particular, C<GP>s
580 can probably be changed with minimal compatibility impact (probably nothing
581 outside of the core, or even outside of F<gv.c> allocates them), but they
582 probably aren't allocated/deallocated often enough for a speed saving. Whereas
583 C<MAGIC> is allocated/deallocated more often, but in turn, is also something
584 more externally visible, so changing the rules here may bite external code.
588 Several SV body structs are now the same size, notably PVMG and PVGV, PVAV and
589 PVHV, and PVCV and PVFM. It should be possible to allocate and return same
590 sized bodies from the same actual arena, rather than maintaining one arena for
591 each. This could save 4-6K per thread, of memory no longer tied up in the
592 not-yet-allocated part of an arena.
595 =head1 Tasks that need a knowledge of XS
597 These tasks would need C knowledge, and roughly the level of knowledge of
598 the perl API that comes from writing modules that use XS to interface to
601 =head2 Write an XS cookbook
603 Create pod/perlxscookbook.pod with short, task-focused 'recipes' in XS that
604 demonstrate common tasks and good practices. (Some of these might be
605 extracted from perlguts.) The target audience should be XS novices, who need
606 more examples than perlguts but something less overwhelming than perlapi.
607 Recipes should provide "one pretty good way to do it" instead of TIMTOWTDI.
609 Rather than focusing on interfacing Perl to C libraries, such a cookbook
610 should probably focus on how to optimize Perl routines by re-writing them
611 in XS. This will likely be more motivating to those who mostly work in
612 Perl but are looking to take the next step into XS.
614 Deconstructing and explaining some simpler XS modules could be one way to
615 bootstrap a cookbook. (List::Util? Class::XSAccessor? Tree::Ternary_XS?)
616 Another option could be deconstructing the implementation of some simpler
619 =head2 Allow XSUBs to inline themselves as OPs
621 For a simple XSUB, often the subroutine dispatch takes more time than the
622 XSUB itself. The tokeniser already has the ability to inline constant
623 subroutines - it would be good to provide a way to inline other subroutines.
625 Specifically, simplest approach looks to be to allow an XSUB to provide an
626 alternative implementation of itself as a custom OP. A new flag bit in
627 C<CvFLAGS()> would signal to the peephole optimiser to take an optree
630 b <@> leave[1 ref] vKP/REFC ->(end)
632 2 <;> nextstate(main 1 -e:1) v:{ ->3
633 a <2> sassign vKS/2 ->b
634 8 <1> entersub[t2] sKS/TARG,1 ->9
637 4 <$> const(IV 1) sM ->5
638 6 <1> rv2av[t1] lKM/1 ->7
640 - <1> ex-rv2cv sK ->-
641 7 <$> gv(*x) s/EARLYCV ->8
642 - <1> ex-rv2sv sKRM*/1 ->a
645 perform the symbol table lookup of C<rv2cv> and C<gv(*x)>, locate the
646 pointer to the custom OP that provides the direct implementation, and re-
647 write the optree something like:
649 b <@> leave[1 ref] vKP/REFC ->(end)
651 2 <;> nextstate(main 1 -e:1) v:{ ->3
652 a <2> sassign vKS/2 ->b
656 4 <$> const(IV 1) sM ->5
657 6 <1> rv2av[t1] lKM/1 ->7
659 - <1> ex-rv2cv sK ->-
660 - <$> ex-gv(*x) s/EARLYCV ->7
661 - <1> ex-rv2sv sKRM*/1 ->a
664 I<i.e.> the C<gv(*)> OP has been nulled and spliced out of the execution
665 path, and the C<entersub> OP has been replaced by the custom op.
667 This approach should provide a measurable speed up to simple XSUBs inside
668 tight loops. Initially one would have to write the OP alternative
669 implementation by hand, but it's likely that this should be reasonably
670 straightforward for the type of XSUB that would benefit the most. Longer
671 term, once the run-time implementation is proven, it should be possible to
672 progressively update ExtUtils::ParseXS to generate OP implementations for
675 =head2 Remove the use of SVs as temporaries in dump.c
677 F<dump.c> contains debugging routines to dump out the contains of perl data
678 structures, such as C<SV>s, C<AV>s and C<HV>s. Currently, the dumping code
679 B<uses> C<SV>s for its temporary buffers, which was a logical initial
680 implementation choice, as they provide ready made memory handling.
682 However, they also lead to a lot of confusion when it happens that what you're
683 trying to debug is seen by the code in F<dump.c>, correctly or incorrectly, as
684 a temporary scalar it can use for a temporary buffer. It's also not possible
685 to dump scalars before the interpreter is properly set up, such as during
686 ithreads cloning. It would be good to progressively replace the use of scalars
687 as string accumulation buffers with something much simpler, directly allocated
688 by C<malloc>. The F<dump.c> code is (or should be) only producing 7 bit
689 US-ASCII, so output character sets are not an issue.
691 Producing and proving an internal simple buffer allocation would make it easier
692 to re-write the internals of the PerlIO subsystem to avoid using C<SV>s for
693 B<its> buffers, use of which can cause problems similar to those of F<dump.c>,
696 =head2 safely supporting POSIX SA_SIGINFO
698 Some years ago Jarkko supplied patches to provide support for the POSIX
699 SA_SIGINFO feature in Perl, passing the extra data to the Perl signal handler.
701 Unfortunately, it only works with "unsafe" signals, because under safe
702 signals, by the time Perl gets to run the signal handler, the extra
703 information has been lost. Moreover, it's not easy to store it somewhere,
704 as you can't call mutexs, or do anything else fancy, from inside a signal
707 So it strikes me that we could provide safe SA_SIGINFO support
713 Provide global variables for two file descriptors
717 When the first request is made via C<sigaction> for C<SA_SIGINFO>, create a
718 pipe, store the reader in one, the writer in the other
722 In the "safe" signal handler (C<Perl_csighandler()>/C<S_raise_signal()>), if
723 the C<siginfo_t> pointer non-C<NULL>, and the writer file handle is open,
729 serialise signal number, C<struct siginfo_t> (or at least the parts we care
730 about) into a small auto char buff
734 C<write()> that (non-blocking) to the writer fd
740 if it writes 100%, flag the signal in a counter of "signals on the pipe" akin
741 to the current per-signal-number counts
745 if it writes 0%, assume the pipe is full. Flag the data as lost?
749 if it writes partially, croak a panic, as your OS is broken.
757 in the regular C<PERL_ASYNC_CHECK()> processing, if there are "signals on
758 the pipe", read the data out, deserialise, build the Perl structures on
759 the stack (code in C<Perl_sighandler()>, the "unsafe" handler), and call as
764 I think that this gets us decent C<SA_SIGINFO> support, without the current risk
765 of running Perl code inside the signal handler context. (With all the dangers
766 of things like C<malloc> corruption that that currently offers us)
768 For more information see the thread starting with this message:
769 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-03/msg00305.html
771 =head2 autovivification
773 Make all autovivification consistent w.r.t LVALUE/RVALUE and strict/no strict;
775 This task is incremental - even a little bit of work on it will help.
777 =head2 Unicode in Filenames
779 chdir, chmod, chown, chroot, exec, glob, link, lstat, mkdir, open,
780 opendir, qx, readdir, readlink, rename, rmdir, stat, symlink, sysopen,
781 system, truncate, unlink, utime, -X. All these could potentially accept
782 Unicode filenames either as input or output (and in the case of system
783 and qx Unicode in general, as input or output to/from the shell).
784 Whether a filesystem - an operating system pair understands Unicode in
787 Known combinations that have some level of understanding include
788 Microsoft NTFS, Apple HFS+ (In Mac OS 9 and X) and Apple UFS (in Mac
789 OS X), NFS v4 is rumored to be Unicode, and of course Plan 9. How to
790 create Unicode filenames, what forms of Unicode are accepted and used
791 (UCS-2, UTF-16, UTF-8), what (if any) is the normalization form used,
792 and so on, varies. Finding the right level of interfacing to Perl
793 requires some thought. Remember that an OS does not implicate a
796 (The Windows -C command flag "wide API support" has been at least
797 temporarily retired in 5.8.1, and the -C has been repurposed, see
800 Most probably the right way to do this would be this:
801 L</"Virtualize operating system access">.
803 =head2 Unicode in %ENV
805 Currently the %ENV entries are always byte strings.
806 See L</"Virtualize operating system access">.
808 =head2 Unicode and glob()
810 Currently glob patterns and filenames returned from File::Glob::glob()
811 are always byte strings. See L</"Virtualize operating system access">.
813 =head2 use less 'memory'
815 Investigate trade offs to switch out perl's choices on memory usage.
816 Particularly perl should be able to give memory back.
818 This task is incremental - even a little bit of work on it will help.
820 =head2 Re-implement C<:unique> in a way that is actually thread-safe
822 The old implementation made bad assumptions on several levels. A good 90%
823 solution might be just to make C<:unique> work to share the string buffer
824 of SvPVs. That way large constant strings can be shared between ithreads,
825 such as the configuration information in F<Config>.
827 =head2 Make tainting consistent
829 Tainting would be easier to use if it didn't take documented shortcuts and
830 allow taint to "leak" everywhere within an expression.
832 =head2 readpipe(LIST)
834 system() accepts a LIST syntax (and a PROGRAM LIST syntax) to avoid
835 running a shell. readpipe() (the function behind qx//) could be similarly
838 =head2 Audit the code for destruction ordering assumptions
842 /* Need to check SvMAGICAL, as during global destruction it may be that
843 AvARYLEN(av) has been freed before av, and hence the SvANY() pointer
844 is now part of the linked list of SV heads, rather than pointing to
845 the original body. */
846 /* FIXME - audit the code for other bugs like this one. */
848 adding the C<SvMAGICAL> check to
850 if (AvARYLEN(av) && SvMAGICAL(AvARYLEN(av))) {
851 MAGIC *mg = mg_find (AvARYLEN(av), PERL_MAGIC_arylen);
853 Go through the core and look for similar assumptions that SVs have particular
854 types, as all bets are off during global destruction.
856 =head2 Extend PerlIO and PerlIO::Scalar
858 PerlIO::Scalar doesn't know how to truncate(). Implementing this
859 would require extending the PerlIO vtable.
861 Similarly the PerlIO vtable doesn't know about formats (write()), or
862 about stat(), or chmod()/chown(), utime(), or flock().
864 (For PerlIO::Scalar it's hard to see what e.g. mode bits or ownership
867 PerlIO doesn't do directories or symlinks, either: mkdir(), rmdir(),
868 opendir(), closedir(), seekdir(), rewinddir(), glob(); symlink(),
871 See also L</"Virtualize operating system access">.
873 =head2 -C on the #! line
875 It should be possible to make -C work correctly if found on the #! line,
876 given that all perl command line options are strict ASCII, and -C changes
877 only the interpretation of non-ASCII characters, and not for the script file
878 handle. To make it work needs some investigation of the ordering of function
879 calls during startup, and (by implication) a bit of tweaking of that order.
881 =head2 Organize error messages
883 Perl's diagnostics (error messages, see L<perldiag>) could use
884 reorganizing and formalizing so that each error message has its
885 stable-for-all-eternity unique id, categorized by severity, type, and
886 subsystem. (The error messages would be listed in a datafile outside
887 of the Perl source code, and the source code would only refer to the
888 messages by the id.) This clean-up and regularizing should apply
889 for all croak() messages.
891 This would enable all sorts of things: easier translation/localization
892 of the messages (though please do keep in mind the caveats of
893 L<Locale::Maketext> about too straightforward approaches to
894 translation), filtering by severity, and instead of grepping for a
895 particular error message one could look for a stable error id. (Of
896 course, changing the error messages by default would break all the
897 existing software depending on some particular error message...)
899 This kind of functionality is known as I<message catalogs>. Look for
900 inspiration for example in the catgets() system, possibly even use it
901 if available-- but B<only> if available, all platforms will B<not>
904 For the really pure at heart, consider extending this item to cover
905 also the warning messages (see L<perllexwarn>, C<warnings.pl>).
907 =head1 Tasks that need a knowledge of the interpreter
909 These tasks would need C knowledge, and knowledge of how the interpreter works,
910 or a willingness to learn.
912 =head2 forbid labels with keyword names
914 Currently C<goto keyword> "computes" the label value:
916 $ perl -e 'goto print'
917 Can't find label 1 at -e line 1.
919 It is controversial if the right way to avoid the confusion is to forbid
920 labels with keyword names, or if it would be better to always treat
921 bareword expressions after a "goto" as a label and never as a keyword.
923 =head2 truncate() prototype
925 The prototype of truncate() is currently C<$$>. It should probably
926 be C<*$> instead. (This is changed in F<opcode.pl>)
928 =head2 decapsulation of smart match argument
930 Currently C<$foo ~~ $object> will die with the message "Smart matching a
931 non-overloaded object breaks encapsulation". It would be nice to allow
932 to bypass this by using explictly the syntax C<$foo ~~ %$object> or
935 =head2 error reporting of [$a ; $b]
937 Using C<;> inside brackets is a syntax error, and we don't propose to change
938 that by giving it any meaning. However, it's not reported very helpfully:
940 $ perl -e '$a = [$b; $c];'
941 syntax error at -e line 1, near "$b;"
942 syntax error at -e line 1, near "$c]"
943 Execution of -e aborted due to compilation errors.
945 It should be possible to hook into the tokeniser or the lexer, so that when a
946 C<;> is parsed where it is not legal as a statement terminator (ie inside
947 C<{}> used as a hashref, C<[]> or C<()>) it issues an error something like
948 I<';' isn't legal inside an expression - if you need multiple statements use a
949 do {...} block>. See the thread starting at
950 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-09/msg00573.html
952 =head2 lexicals used only once
956 $ perl -we '$pie = 42'
957 Name "main::pie" used only once: possible typo at -e line 1.
961 $ perl -we 'my $pie = 42'
963 Logically all lexicals used only once should warn, if the user asks for
964 warnings. An unworked RT ticket (#5087) has been open for almost seven
965 years for this discrepancy.
969 The handling of Unicode is unclean in many places. For example, the regexp
970 engine matches in Unicode semantics whenever the string or the pattern is
971 flagged as UTF-8, but that should not be dependent on an internal storage
972 detail of the string.
974 =head2 Properly Unicode safe tokeniser and pads.
976 The tokeniser isn't actually very UTF-8 clean. C<use utf8;> is a hack -
977 variable names are stored in stashes as raw bytes, without the utf-8 flag
978 set. The pad API only takes a C<char *> pointer, so that's all bytes too. The
979 tokeniser ignores the UTF-8-ness of C<PL_rsfp>, or any SVs returned from
980 source filters. All this could be fixed.
982 =head2 state variable initialization in list context
984 Currently this is illegal:
986 state ($a, $b) = foo();
988 In Perl 6, C<state ($a) = foo();> and C<(state $a) = foo();> have different
989 semantics, which is tricky to implement in Perl 5 as currently they produce
990 the same opcode trees. The Perl 6 design is firm, so it would be good to
991 implement the necessary code in Perl 5. There are comments in
992 C<Perl_newASSIGNOP()> that show the code paths taken by various assignment
993 constructions involving state variables.
995 =head2 Implement $value ~~ 0 .. $range
997 It would be nice to extend the syntax of the C<~~> operator to also
998 understand numeric (and maybe alphanumeric) ranges.
1000 =head2 A does() built-in
1002 Like ref(), only useful. It would call the C<DOES> method on objects; it
1003 would also tell whether something can be dereferenced as an
1004 array/hash/etc., or used as a regexp, etc.
1005 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-03/msg00481.html>
1007 =head2 Tied filehandles and write() don't mix
1009 There is no method on tied filehandles to allow them to be called back by
1012 =head2 Propagate compilation hints to the debugger
1014 Currently a debugger started with -dE on the command-line doesn't see the
1015 features enabled by -E. More generally hints (C<$^H> and C<%^H>) aren't
1016 propagated to the debugger. Probably it would be a good thing to propagate
1017 hints from the innermost non-C<DB::> scope: this would make code eval'ed
1018 in the debugger see the features (and strictures, etc.) currently in
1021 =head2 Attach/detach debugger from running program
1023 The old perltodo notes "With C<gdb>, you can attach the debugger to a running
1024 program if you pass the process ID. It would be good to do this with the Perl
1025 debugger on a running Perl program, although I'm not sure how it would be
1026 done." ssh and screen do this with named pipes in /tmp. Maybe we can too.
1028 =head2 LVALUE functions for lists
1030 The old perltodo notes that lvalue functions don't work for list or hash
1031 slices. This would be good to fix.
1033 =head2 regexp optimiser optional
1035 The regexp optimiser is not optional. It should configurable to be, to allow
1036 its performance to be measured, and its bugs to be easily demonstrated.
1038 =head2 C</w> regex modifier
1040 That flag would enable to match whole words, and also to interpolate
1041 arrays as alternations. With it, C</P/w> would be roughly equivalent to:
1043 do { local $"='|'; /\b(?:P)\b/ }
1045 See L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-01/msg00400.html>
1048 =head2 optional optimizer
1050 Make the peephole optimizer optional. Currently it performs two tasks as
1051 it walks the optree - genuine peephole optimisations, and necessary fixups of
1052 ops. It would be good to find an efficient way to switch out the
1053 optimisations whilst keeping the fixups.
1055 =head2 You WANT *how* many
1057 Currently contexts are void, scalar and list. split has a special mechanism in
1058 place to pass in the number of return values wanted. It would be useful to
1059 have a general mechanism for this, backwards compatible and little speed hit.
1060 This would allow proposals such as short circuiting sort to be implemented
1061 as a module on CPAN.
1063 =head2 lexical aliases
1065 Allow lexical aliases (maybe via the syntax C<my \$alias = \$foo>.
1067 =head2 entersub XS vs Perl
1069 At the moment pp_entersub is huge, and has code to deal with entering both
1070 perl and XS subroutines. Subroutine implementations rarely change between
1071 perl and XS at run time, so investigate using 2 ops to enter subs (one for
1072 XS, one for perl) and swap between if a sub is redefined.
1076 Self-ties are currently illegal because they caused too many segfaults. Maybe
1077 the causes of these could be tracked down and self-ties on all types
1080 =head2 Optimize away @_
1082 The old perltodo notes "Look at the "reification" code in C<av.c>".
1084 =head2 Virtualize operating system access
1086 Implement a set of "vtables" that virtualizes operating system access
1087 (open(), mkdir(), unlink(), readdir(), getenv(), etc.) At the very
1088 least these interfaces should take SVs as "name" arguments instead of
1089 bare char pointers; probably the most flexible and extensible way
1090 would be for the Perl-facing interfaces to accept HVs. The system
1091 needs to be per-operating-system and per-file-system
1092 hookable/filterable, preferably both from XS and Perl level
1093 (L<perlport/"Files and Filesystems"> is good reading at this point,
1094 in fact, all of L<perlport> is.)
1096 This has actually already been implemented (but only for Win32),
1097 take a look at F<iperlsys.h> and F<win32/perlhost.h>. While all Win32
1098 variants go through a set of "vtables" for operating system access,
1099 non-Win32 systems currently go straight for the POSIX/Unix-style
1100 system/library call. Similar system as for Win32 should be
1101 implemented for all platforms. The existing Win32 implementation
1102 probably does not need to survive alongside this proposed new
1103 implementation, the approaches could be merged.
1105 What would this give us? One often-asked-for feature this would
1106 enable is using Unicode for filenames, and other "names" like %ENV,
1107 usernames, hostnames, and so forth.
1108 (See L<perlunicode/"When Unicode Does Not Happen">.)
1110 But this kind of virtualization would also allow for things like
1111 virtual filesystems, virtual networks, and "sandboxes" (though as long
1112 as dynamic loading of random object code is allowed, not very safe
1113 sandboxes since external code of course know not of Perl's vtables).
1114 An example of a smaller "sandbox" is that this feature can be used to
1115 implement per-thread working directories: Win32 already does this.
1117 See also L</"Extend PerlIO and PerlIO::Scalar">.
1119 =head2 Investigate PADTMP hash pessimisation
1121 The peephole optimiser converts constants used for hash key lookups to shared
1122 hash key scalars. Under ithreads, something is undoing this work.
1123 See http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-09/msg00793.html
1125 =head2 Store the current pad in the OP slab allocator
1128 I hope that I got that "current pad" part correct
1130 Currently we leak ops in various cases of parse failure. I suggested that we
1131 could solve this by always using the op slab allocator, and walking it to
1132 free ops. Dave comments that as some ops are already freed during optree
1133 creation one would have to mark which ops are freed, and not double free them
1134 when walking the slab. He notes that one problem with this is that for some ops
1135 you have to know which pad was current at the time of allocation, which does
1136 change. I suggested storing a pointer to the current pad in the memory allocated
1137 for the slab, and swapping to a new slab each time the pad changes. Dave thinks
1138 that this would work.
1140 =head2 repack the optree
1142 Repacking the optree after execution order is determined could allow
1143 removal of NULL ops, and optimal ordering of OPs with respect to cache-line
1144 filling. The slab allocator could be reused for this purpose. I think that
1145 the best way to do this is to make it an optional step just before the
1146 completed optree is attached to anything else, and to use the slab allocator
1147 unchanged, so that freeing ops is identical whether or not this step runs.
1148 Note that the slab allocator allocates ops downwards in memory, so one would
1149 have to actually "allocate" the ops in reverse-execution order to get them
1150 contiguous in memory in execution order.
1152 See http://www.nntp.perl.org/group/perl.perl5.porters/2007/12/msg131975.html
1154 Note that running this copy, and then freeing all the old location ops would
1155 cause their slabs to be freed, which would eliminate possible memory wastage if
1156 the previous suggestion is implemented, and we swap slabs more frequently.
1158 =head2 eliminate incorrect line numbers in warnings
1166 } elsif ($undef == 0) {
1169 used to produce this output:
1171 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1172 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1174 where the line of the second warning was misreported - it should be line 5.
1175 Rafael fixed this - the problem arose because there was no nextstate OP
1176 between the execution of the C<if> and the C<elsif>, hence C<PL_curcop> still
1177 reports that the currently executing line is line 4. The solution was to inject
1178 a nextstate OPs for each C<elsif>, although it turned out that the nextstate
1179 OP needed to be a nulled OP, rather than a live nextstate OP, else other line
1180 numbers became misreported. (Jenga!)
1182 The problem is more general than C<elsif> (although the C<elsif> case is the
1183 most common and the most confusing). Ideally this code
1193 would produce this output
1195 Use of uninitialized value $undef in addition (+) at wrong.pl line 4.
1196 Use of uninitialized value $undef in addition (+) at wrong.pl line 7.
1198 (rather than lines 4 and 5), but this would seem to require every OP to carry
1199 (at least) line number information.
1201 What might work is to have an optional line number in memory just before the
1202 BASEOP structure, with a flag bit in the op to say whether it's present.
1203 Initially during compile every OP would carry its line number. Then add a late
1204 pass to the optimiser (potentially combined with L</repack the optree>) which
1205 looks at the two ops on every edge of the graph of the execution path. If
1206 the line number changes, flags the destination OP with this information.
1207 Once all paths are traced, replace every op with the flag with a
1208 nextstate-light op (that just updates C<PL_curcop>), which in turn then passes
1209 control on to the true op. All ops would then be replaced by variants that
1210 do not store the line number. (Which, logically, why it would work best in
1211 conjunction with L</repack the optree>, as that is already copying/reallocating
1214 (Although I should note that we're not certain that doing this for the general
1217 =head2 optimize tail-calls
1219 Tail-calls present an opportunity for broadly applicable optimization;
1220 anywhere that C<< return foo(...) >> is called, the outer return can
1221 be replaced by a goto, and foo will return directly to the outer
1222 caller, saving (conservatively) 25% of perl's call&return cost, which
1223 is relatively higher than in C. The scheme language is known to do
1224 this heavily. B::Concise provides good insight into where this
1225 optimization is possible, ie anywhere entersub,leavesub op-sequence
1228 perl -MO=Concise,-exec,a,b,-main -e 'sub a{ 1 }; sub b {a()}; b(2)'
1230 Bottom line on this is probably a new pp_tailcall function which
1231 combines the code in pp_entersub, pp_leavesub. This should probably
1232 be done 1st in XS, and using B::Generate to patch the new OP into the
1235 =head2 Add C<00dddd>
1237 It has been proposed that octal constants be specifiable through the syntax
1238 C<0oddddd>, parallel to the existing construct to specify hex constants
1243 Tasks that will get your name mentioned in the description of the "Highlights
1246 =head2 make ithreads more robust
1248 Generally make ithreads more robust. See also L</iCOW>
1250 This task is incremental - even a little bit of work on it will help, and
1251 will be greatly appreciated.
1253 One bit would be to determine how to clone directory handles on systems
1254 without a C<fchdir> function (in sv.c:Perl_dirp_dup).
1256 Fix Perl_sv_dup, et al so that threads can return objects.
1260 Sarathy and Arthur have a proposal for an improved Copy On Write which
1261 specifically will be able to COW new ithreads. If this can be implemented
1262 it would be a good thing.
1264 =head2 (?{...}) closures in regexps
1266 Fix (or rewrite) the implementation of the C</(?{...})/> closures.
1268 =head2 Add class set operations to regexp engine
1270 Apparently these are quite useful. Anyway, Jeffery Friedl wants them.
1272 demerphq has this on his todo list, but right at the bottom.
1275 =head1 Tasks for microperl
1278 [ Each and every one of these may be obsolete, but they were listed
1279 in the old Todo.micro file]
1282 =head2 make creating uconfig.sh automatic
1284 =head2 make creating Makefile.micro automatic
1286 =head2 do away with fork/exec/wait?
1288 (system, popen should be enough?)
1290 =head2 some of the uconfig.sh really needs to be probed (using cc) in buildtime:
1292 (uConfigure? :-) native datatype widths and endianness come to mind