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 Remove macperl references from tests
31 MacPerl is gone. The tests don't need to be there.
33 =head2 Remove duplication of test setup.
35 Schwern notes, that there's duplication of code - lots and lots of tests have
36 some variation on the big block of C<$Is_Foo> checks. We can safely put this
37 into a file, change it to build an C<%Is> hash and require it. Maybe just put
38 it into F<test.pl>. Throw in the handy tainting subroutines.
40 =head2 POD -E<gt> HTML conversion in the core still sucks
42 Which is crazy given just how simple POD purports to be, and how simple HTML
43 can be. It's not actually I<as> simple as it sounds, particularly with the
44 flexibility POD allows for C<=item>, but it would be good to improve the
45 visual appeal of the HTML generated, and to avoid it having any validation
46 errors. See also L</make HTML install work>, as the layout of installation tree
47 is needed to improve the cross-linking.
49 The addition of C<Pod::Simple> and its related modules may make this task
52 =head2 Make ExtUtils::ParseXS use strict;
54 F<lib/ExtUtils/ParseXS.pm> contains this line
56 # use strict; # One of these days...
58 Simply uncomment it, and fix all the resulting issues :-)
60 The more practical approach, to break the task down into manageable chunks, is
61 to work your way though the code from bottom to top, or if necessary adding
62 extra C<{ ... }> blocks, and turning on strict within them.
64 =head2 Make Schwern poorer
66 We should have tests for everything. When all the core's modules are tested,
67 Schwern has promised to donate to $500 to TPF. We may need volunteers to
68 hold him upside down and shake vigorously in order to actually extract the
71 =head2 Improve the coverage of the core tests
73 Use Devel::Cover to ascertain the core modules's test coverage, then add
74 tests that are currently missing.
78 A full test suite for the B module would be nice.
80 =head2 A decent benchmark
82 C<perlbench> seems impervious to any recent changes made to the perl core. It
83 would be useful to have a reasonable general benchmarking suite that roughly
84 represented what current perl programs do, and measurably reported whether
85 tweaks to the core improve, degrade or don't really affect performance, to
86 guide people attempting to optimise the guts of perl. Gisle would welcome
87 new tests for perlbench.
89 =head2 fix tainting bugs
91 Fix the bugs revealed by running the test suite with the C<-t> switch (via
92 C<make test.taintwarn>).
94 =head2 Dual life everything
96 As part of the "dists" plan, anything that doesn't belong in the smallest perl
97 distribution needs to be dual lifed. Anything else can be too. Figure out what
98 changes would be needed to package that module and its tests up for CPAN, and
99 do so. Test it with older perl releases, and fix the problems you find.
101 To make a minimal perl distribution, it's useful to look at
102 F<t/lib/commonsense.t>.
104 =head2 Move dual-life pod/*.PL into ext
106 Nearly all the dual-life modules have been moved to F<ext>. However, we
107 still need to move F<pod/*.PL> into their respective directories
108 in F<ext/>. They're referenced by (at least) C<plextract> in F<Makefile.SH>
109 and C<utils> in F<win32/Makefile> and F<win32/makefile.ml>, and listed
110 explicitly in F<win32/pod.mak>, F<vms/descrip_mms.template> and F<utils.lst>
112 =head2 POSIX memory footprint
114 Ilya observed that use POSIX; eats memory like there's no tomorrow, and at
115 various times worked to cut it down. There is probably still fat to cut out -
116 for example POSIX passes Exporter some very memory hungry data structures.
118 =head2 embed.pl/makedef.pl
120 There is a script F<embed.pl> that generates several header files to prefix
121 all of Perl's symbols in a consistent way, to provide some semblance of
122 namespace support in C<C>. Functions are declared in F<embed.fnc>, variables
123 in F<interpvar.h>. Quite a few of the functions and variables
124 are conditionally declared there, using C<#ifdef>. However, F<embed.pl>
125 doesn't understand the C macros, so the rules about which symbols are present
126 when is duplicated in F<makedef.pl>. Writing things twice is bad, m'kay.
127 It would be good to teach C<embed.pl> to understand the conditional
128 compilation, and hence remove the duplication, and the mistakes it has caused.
130 =head2 use strict; and AutoLoad
132 Currently if you write
135 use AutoLoader 'AUTOLOAD';
140 print join (' ', No, strict, here), "!\n";
143 then C<use strict;> isn't in force within the autoloaded subroutines. It would
144 be more consistent (and less surprising) to arrange for all lexical pragmas
145 in force at the __END__ block to be in force within each autoloaded subroutine.
147 There's a similar problem with SelfLoader.
149 =head2 profile installman
151 The F<installman> script is slow. All it is doing text processing, which we're
152 told is something Perl is good at. So it would be nice to know what it is doing
153 that is taking so much CPU, and where possible address it.
156 =head1 Tasks that need a little sysadmin-type knowledge
158 Or if you prefer, tasks that you would learn from, and broaden your skills
161 =head2 make HTML install work
163 There is an C<installhtml> target in the Makefile. It's marked as
164 "experimental". It would be good to get this tested, make it work reliably, and
165 remove the "experimental" tag. This would include
171 Checking that cross linking between various parts of the documentation works.
172 In particular that links work between the modules (files with POD in F<lib/>)
173 and the core documentation (files in F<pod/>)
177 Work out how to split C<perlfunc> into chunks, preferably one per function
178 group, preferably with general case code that could be used elsewhere.
179 Challenges here are correctly identifying the groups of functions that go
180 together, and making the right named external cross-links point to the right
181 page. Things to be aware of are C<-X>, groups such as C<getpwnam> to
182 C<endservent>, two or more C<=items> giving the different parameter lists, such
185 =item substr EXPR,OFFSET,LENGTH,REPLACEMENT
186 =item substr EXPR,OFFSET,LENGTH
187 =item substr EXPR,OFFSET
189 and different parameter lists having different meanings. (eg C<select>)
193 =head2 compressed man pages
195 Be able to install them. This would probably need a configure test to see how
196 the system does compressed man pages (same directory/different directory?
197 same filename/different filename), as well as tweaking the F<installman> script
198 to compress as necessary.
200 =head2 Add a code coverage target to the Makefile
202 Make it easy for anyone to run Devel::Cover on the core's tests. The steps
203 to do this manually are roughly
209 do a normal C<Configure>, but include Devel::Cover as a module to install
210 (see F<INSTALL> for how to do this)
218 cd t; HARNESS_PERL_SWITCHES=-MDevel::Cover ./perl -I../lib harness
222 Process the resulting Devel::Cover database
226 This just give you the coverage of the F<.pm>s. To also get the C level
233 Additionally tell C<Configure> to use the appropriate C compiler flags for
240 (instead of C<make perl>)
244 After running the tests run C<gcov> to generate all the F<.gcov> files.
245 (Including down in the subdirectories of F<ext/>
249 (From the top level perl directory) run C<gcov2perl> on all the C<.gcov> files
250 to get their stats into the cover_db directory.
254 Then process the Devel::Cover database
258 It would be good to add a single switch to C<Configure> to specify that you
259 wanted to perform perl level coverage, and another to specify C level
260 coverage, and have C<Configure> and the F<Makefile> do all the right things
263 =head2 Make Config.pm cope with differences between built and installed perl
265 Quite often vendors ship a perl binary compiled with their (pay-for)
266 compilers. People install a free compiler, such as gcc. To work out how to
267 build extensions, Perl interrogates C<%Config>, so in this situation
268 C<%Config> describes compilers that aren't there, and extension building
269 fails. This forces people into choosing between re-compiling perl themselves
270 using the compiler they have, or only using modules that the vendor ships.
272 It would be good to find a way teach C<Config.pm> about the installation setup,
273 possibly involving probing at install time or later, so that the C<%Config> in
274 a binary distribution better describes the installed machine, when the
275 installed machine differs from the build machine in some significant way.
277 =head2 linker specification files
279 Some platforms mandate that you provide a list of a shared library's external
280 symbols to the linker, so the core already has the infrastructure in place to
281 do this for generating shared perl libraries. My understanding is that the
282 GNU toolchain can accept an optional linker specification file, and restrict
283 visibility just to symbols declared in that file. It would be good to extend
284 F<makedef.pl> to support this format, and to provide a means within
285 C<Configure> to enable it. This would allow Unix users to test that the
286 export list is correct, and to build a perl that does not pollute the global
287 namespace with private symbols.
289 =head2 Cross-compile support
291 Currently C<Configure> understands C<-Dusecrosscompile> option. This option
292 arranges for building C<miniperl> for TARGET machine, so this C<miniperl> is
293 assumed then to be copied to TARGET machine and used as a replacement of full
296 This could be done little differently. Namely C<miniperl> should be built for
297 HOST and then full C<perl> with extensions should be compiled for TARGET.
298 This, however, might require extra trickery for %Config: we have one config
299 first for HOST and then another for TARGET. Tools like MakeMaker will be
300 mightily confused. Having around two different types of executables and
301 libraries (HOST and TARGET) makes life interesting for Makefiles and
302 shell (and Perl) scripts. There is $Config{run}, normally empty, which
303 can be used as an execution wrapper. Also note that in some
304 cross-compilation/execution environments the HOST and the TARGET do
305 not see the same filesystem(s), the $Config{run} may need to do some
306 file/directory copying back and forth.
310 Make F<pod/roffitall> be updated by F<pod/buildtoc>.
312 =head2 Split "linker" from "compiler"
314 Right now, Configure probes for two commands, and sets two variables:
318 =item * C<cc> (in F<cc.U>)
320 This variable holds the name of a command to execute a C compiler which
321 can resolve multiple global references that happen to have the same
322 name. Usual values are F<cc> and F<gcc>.
323 Fervent ANSI compilers may be called F<c89>. AIX has F<xlc>.
325 =item * C<ld> (in F<dlsrc.U>)
327 This variable indicates the program to be used to link
328 libraries for dynamic loading. On some systems, it is F<ld>.
329 On ELF systems, it should be C<$cc>. Mostly, we'll try to respect
330 the hint file setting.
334 There is an implicit historical assumption from around Perl5.000alpha
335 something, that C<$cc> is also the correct command for linking object files
336 together to make an executable. This may be true on Unix, but it's not true
337 on other platforms, and there are a maze of work arounds in other places (such
338 as F<Makefile.SH>) to cope with this.
340 Ideally, we should create a new variable to hold the name of the executable
341 linker program, probe for it in F<Configure>, and centralise all the special
342 case logic there or in hints files.
344 A small bikeshed issue remains - what to call it, given that C<$ld> is already
345 taken (arguably for the wrong thing now, but on SunOS 4.1 it is the command
346 for creating dynamically-loadable modules) and C<$link> could be confused with
347 the Unix command line executable of the same name, which does something
348 completely different. Andy Dougherty makes the counter argument "In parrot, I
349 tried to call the command used to link object files and libraries into an
350 executable F<link>, since that's what my vaguely-remembered DOS and VMS
351 experience suggested. I don't think any real confusion has ensued, so it's
352 probably a reasonable name for perl5 to use."
354 "Alas, I've always worried that introducing it would make things worse,
355 since now the module building utilities would have to look for
356 C<$Config{link}> and institute a fall-back plan if it weren't found."
357 Although I can see that as confusing, given that C<$Config{d_link}> is true
358 when (hard) links are available.
360 =head2 Configure Windows using PowerShell
362 Currently, Windows uses hard-coded config files based to build the
363 config.h for compiling Perl. Makefiles are also hard-coded and need to be
364 hand edited prior to building Perl. While this makes it easy to create a perl.exe
365 that works across multiple Windows versions, being able to accurately
366 configure a perl.exe for a specific Windows versions and VS C++ would be
367 a nice enhancement. With PowerShell available on Windows XP and up, this
368 may now be possible. Step 1 might be to investigate whether this is possible
369 and use this to clean up our current makefile situation. Step 2 would be to
370 see if there would be a way to use our existing metaconfig units to configure a
371 Windows Perl or whether we go in a separate direction and make it so. Of
372 course, we all know what step 3 is.
374 =head2 decouple -g and -DDEBUGGING
376 Currently F<Configure> automatically adds C<-DDEBUGGING> to the C compiler
377 flags if it spots C<-g> in the optimiser flags. The pre-processor directive
378 C<DEBUGGING> enables F<perl>'s command line C<-D> options, but in the process
379 makes F<perl> slower. It would be good to disentangle this logic, so that
380 C-level debugging with C<-g> and Perl level debugging with C<-D> can easily
381 be enabled independently.
383 =head1 Tasks that need a little C knowledge
385 These tasks would need a little C knowledge, but don't need any specific
386 background or experience with XS, or how the Perl interpreter works
388 =head2 Weed out needless PERL_UNUSED_ARG
390 The C code uses the macro C<PERL_UNUSED_ARG> to stop compilers warning about
391 unused arguments. Often the arguments can't be removed, as there is an
392 external constraint that determines the prototype of the function, so this
393 approach is valid. However, there are some cases where C<PERL_UNUSED_ARG>
394 could be removed. Specifically
400 The prototypes of (nearly all) static functions can be changed
404 Unused arguments generated by short cut macros are wasteful - the short cut
405 macro used can be changed.
409 =head2 Modernize the order of directories in @INC
411 The way @INC is laid out by default, one cannot upgrade core (dual-life)
412 modules without overwriting files. This causes problems for binary
413 package builders. One possible proposal is laid out in this
415 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2002-04/msg02380.html>.
419 Natively 64-bit systems need neither -Duse64bitint nor -Duse64bitall.
420 On these systems, it might be the default compilation mode, and there
421 is currently no guarantee that passing no use64bitall option to the
422 Configure process will build a 32bit perl. Implementing -Duse32bit*
423 options would be nice for perl 5.12.
425 =head2 Profile Perl - am I hot or not?
427 The Perl source code is stable enough that it makes sense to profile it,
428 identify and optimise the hotspots. It would be good to measure the
429 performance of the Perl interpreter using free tools such as cachegrind,
430 gprof, and dtrace, and work to reduce the bottlenecks they reveal.
432 As part of this, the idea of F<pp_hot.c> is that it contains the I<hot> ops,
433 the ops that are most commonly used. The idea is that by grouping them, their
434 object code will be adjacent in the executable, so they have a greater chance
435 of already being in the CPU cache (or swapped in) due to being near another op
438 Except that it's not clear if these really are the most commonly used ops. So
439 as part of exercising your skills with coverage and profiling tools you might
440 want to determine what ops I<really> are the most commonly used. And in turn
441 suggest evictions and promotions to achieve a better F<pp_hot.c>.
443 One piece of Perl code that might make a good testbed is F<installman>.
445 =head2 Allocate OPs from arenas
447 Currently all new OP structures are individually malloc()ed and free()d.
448 All C<malloc> implementations have space overheads, and are now as fast as
449 custom allocates so it would both use less memory and less CPU to allocate
450 the various OP structures from arenas. The SV arena code can probably be
453 Note that Configuring perl with C<-Accflags=-DPL_OP_SLAB_ALLOC> will use
454 Perl_Slab_alloc() to pack optrees into a contiguous block, which is
455 probably superior to the use of OP arenas, esp. from a cache locality
456 standpoint. See L<Profile Perl - am I hot or not?>.
458 =head2 Improve win32/wince.c
460 Currently, numerous functions look virtually, if not completely,
461 identical in both C<win32/wince.c> and C<win32/win32.c> files, which can't
464 =head2 Use secure CRT functions when building with VC8 on Win32
466 Visual C++ 2005 (VC++ 8.x) deprecated a number of CRT functions on the basis
467 that they were "unsafe" and introduced differently named secure versions of
468 them as replacements, e.g. instead of writing
470 FILE* f = fopen(__FILE__, "r");
475 errno_t err = fopen_s(&f, __FILE__, "r");
477 Currently, the warnings about these deprecations have been disabled by adding
478 -D_CRT_SECURE_NO_DEPRECATE to the CFLAGS. It would be nice to remove that
479 warning suppressant and actually make use of the new secure CRT functions.
481 There is also a similar issue with POSIX CRT function names like fileno having
482 been deprecated in favour of ISO C++ conformant names like _fileno. These
483 warnings are also currently suppressed by adding -D_CRT_NONSTDC_NO_DEPRECATE. It
484 might be nice to do as Microsoft suggest here too, although, unlike the secure
485 functions issue, there is presumably little or no benefit in this case.
487 =head2 Fix POSIX::access() and chdir() on Win32
489 These functions currently take no account of DACLs and therefore do not behave
490 correctly in situations where access is restricted by DACLs (as opposed to the
491 read-only attribute).
493 Furthermore, POSIX::access() behaves differently for directories having the
494 read-only attribute set depending on what CRT library is being used. For
495 example, the _access() function in the VC6 and VC7 CRTs (wrongly) claim that
496 such directories are not writable, whereas in fact all directories are writable
497 unless access is denied by DACLs. (In the case of directories, the read-only
498 attribute actually only means that the directory cannot be deleted.) This CRT
499 bug is fixed in the VC8 and VC9 CRTs (but, of course, the directory may still
500 not actually be writable if access is indeed denied by DACLs).
502 For the chdir() issue, see ActiveState bug #74552:
503 http://bugs.activestate.com/show_bug.cgi?id=74552
505 Therefore, DACLs should be checked both for consistency across CRTs and for
508 (Note that perl's -w operator should not be modified to check DACLs. It has
509 been written so that it reflects the state of the read-only attribute, even
510 for directories (whatever CRT is being used), for symmetry with chmod().)
512 =head2 strcat(), strcpy(), strncat(), strncpy(), sprintf(), vsprintf()
514 Maybe create a utility that checks after each libperl.a creation that
515 none of the above (nor sprintf(), vsprintf(), or *SHUDDER* gets())
516 ever creep back to libperl.a.
518 nm libperl.a | ./miniperl -alne '$o = $F[0] if /:$/; print "$o $F[1]" if $F[0] eq "U" && $F[1] =~ /^(?:strn?c(?:at|py)|v?sprintf|gets)$/'
520 Note, of course, that this will only tell whether B<your> platform
521 is using those naughty interfaces.
523 =head2 -D_FORTIFY_SOURCE=2, -fstack-protector
525 Recent glibcs support C<-D_FORTIFY_SOURCE=2> and recent gcc
526 (4.1 onwards?) supports C<-fstack-protector>, both of which give
527 protection against various kinds of buffer overflow problems.
528 These should probably be used for compiling Perl whenever available,
529 Configure and/or hints files should be adjusted to probe for the
530 availability of these features and enable them as appropriate.
532 =head2 Arenas for GPs? For MAGIC?
534 C<struct gp> and C<struct magic> are both currently allocated by C<malloc>.
535 It might be a speed or memory saving to change to using arenas. Or it might
536 not. It would need some suitable benchmarking first. In particular, C<GP>s
537 can probably be changed with minimal compatibility impact (probably nothing
538 outside of the core, or even outside of F<gv.c> allocates them), but they
539 probably aren't allocated/deallocated often enough for a speed saving. Whereas
540 C<MAGIC> is allocated/deallocated more often, but in turn, is also something
541 more externally visible, so changing the rules here may bite external code.
545 Several SV body structs are now the same size, notably PVMG and PVGV, PVAV and
546 PVHV, and PVCV and PVFM. It should be possible to allocate and return same
547 sized bodies from the same actual arena, rather than maintaining one arena for
548 each. This could save 4-6K per thread, of memory no longer tied up in the
549 not-yet-allocated part of an arena.
552 =head1 Tasks that need a knowledge of XS
554 These tasks would need C knowledge, and roughly the level of knowledge of
555 the perl API that comes from writing modules that use XS to interface to
558 =head2 Remove the use of SVs as temporaries in dump.c
560 F<dump.c> contains debugging routines to dump out the contains of perl data
561 structures, such as C<SV>s, C<AV>s and C<HV>s. Currently, the dumping code
562 B<uses> C<SV>s for its temporary buffers, which was a logical initial
563 implementation choice, as they provide ready made memory handling.
565 However, they also lead to a lot of confusion when it happens that what you're
566 trying to debug is seen by the code in F<dump.c>, correctly or incorrectly, as
567 a temporary scalar it can use for a temporary buffer. It's also not possible
568 to dump scalars before the interpreter is properly set up, such as during
569 ithreads cloning. It would be good to progressively replace the use of scalars
570 as string accumulation buffers with something much simpler, directly allocated
571 by C<malloc>. The F<dump.c> code is (or should be) only producing 7 bit
572 US-ASCII, so output character sets are not an issue.
574 Producing and proving an internal simple buffer allocation would make it easier
575 to re-write the internals of the PerlIO subsystem to avoid using C<SV>s for
576 B<its> buffers, use of which can cause problems similar to those of F<dump.c>,
579 =head2 safely supporting POSIX SA_SIGINFO
581 Some years ago Jarkko supplied patches to provide support for the POSIX
582 SA_SIGINFO feature in Perl, passing the extra data to the Perl signal handler.
584 Unfortunately, it only works with "unsafe" signals, because under safe
585 signals, by the time Perl gets to run the signal handler, the extra
586 information has been lost. Moreover, it's not easy to store it somewhere,
587 as you can't call mutexs, or do anything else fancy, from inside a signal
590 So it strikes me that we could provide safe SA_SIGINFO support
596 Provide global variables for two file descriptors
600 When the first request is made via C<sigaction> for C<SA_SIGINFO>, create a
601 pipe, store the reader in one, the writer in the other
605 In the "safe" signal handler (C<Perl_csighandler()>/C<S_raise_signal()>), if
606 the C<siginfo_t> pointer non-C<NULL>, and the writer file handle is open,
612 serialise signal number, C<struct siginfo_t> (or at least the parts we care
613 about) into a small auto char buff
617 C<write()> that (non-blocking) to the writer fd
623 if it writes 100%, flag the signal in a counter of "signals on the pipe" akin
624 to the current per-signal-number counts
628 if it writes 0%, assume the pipe is full. Flag the data as lost?
632 if it writes partially, croak a panic, as your OS is broken.
640 in the regular C<PERL_ASYNC_CHECK()> processing, if there are "signals on
641 the pipe", read the data out, deserialise, build the Perl structures on
642 the stack (code in C<Perl_sighandler()>, the "unsafe" handler), and call as
647 I think that this gets us decent C<SA_SIGINFO> support, without the current risk
648 of running Perl code inside the signal handler context. (With all the dangers
649 of things like C<malloc> corruption that that currently offers us)
651 For more information see the thread starting with this message:
652 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-03/msg00305.html
654 =head2 autovivification
656 Make all autovivification consistent w.r.t LVALUE/RVALUE and strict/no strict;
658 This task is incremental - even a little bit of work on it will help.
660 =head2 Unicode in Filenames
662 chdir, chmod, chown, chroot, exec, glob, link, lstat, mkdir, open,
663 opendir, qx, readdir, readlink, rename, rmdir, stat, symlink, sysopen,
664 system, truncate, unlink, utime, -X. All these could potentially accept
665 Unicode filenames either as input or output (and in the case of system
666 and qx Unicode in general, as input or output to/from the shell).
667 Whether a filesystem - an operating system pair understands Unicode in
670 Known combinations that have some level of understanding include
671 Microsoft NTFS, Apple HFS+ (In Mac OS 9 and X) and Apple UFS (in Mac
672 OS X), NFS v4 is rumored to be Unicode, and of course Plan 9. How to
673 create Unicode filenames, what forms of Unicode are accepted and used
674 (UCS-2, UTF-16, UTF-8), what (if any) is the normalization form used,
675 and so on, varies. Finding the right level of interfacing to Perl
676 requires some thought. Remember that an OS does not implicate a
679 (The Windows -C command flag "wide API support" has been at least
680 temporarily retired in 5.8.1, and the -C has been repurposed, see
683 Most probably the right way to do this would be this:
684 L</"Virtualize operating system access">.
686 =head2 Unicode in %ENV
688 Currently the %ENV entries are always byte strings.
689 See L</"Virtualize operating system access">.
691 =head2 Unicode and glob()
693 Currently glob patterns and filenames returned from File::Glob::glob()
694 are always byte strings. See L</"Virtualize operating system access">.
696 =head2 Unicode and lc/uc operators
698 Some built-in operators (C<lc>, C<uc>, etc.) behave differently, based on
699 what the internal encoding of their argument is. That should not be the
700 case. Maybe add a pragma to switch behaviour.
702 =head2 use less 'memory'
704 Investigate trade offs to switch out perl's choices on memory usage.
705 Particularly perl should be able to give memory back.
707 This task is incremental - even a little bit of work on it will help.
709 =head2 Re-implement C<:unique> in a way that is actually thread-safe
711 The old implementation made bad assumptions on several levels. A good 90%
712 solution might be just to make C<:unique> work to share the string buffer
713 of SvPVs. That way large constant strings can be shared between ithreads,
714 such as the configuration information in F<Config>.
716 =head2 Make tainting consistent
718 Tainting would be easier to use if it didn't take documented shortcuts and
719 allow taint to "leak" everywhere within an expression.
721 =head2 readpipe(LIST)
723 system() accepts a LIST syntax (and a PROGRAM LIST syntax) to avoid
724 running a shell. readpipe() (the function behind qx//) could be similarly
727 =head2 Audit the code for destruction ordering assumptions
731 /* Need to check SvMAGICAL, as during global destruction it may be that
732 AvARYLEN(av) has been freed before av, and hence the SvANY() pointer
733 is now part of the linked list of SV heads, rather than pointing to
734 the original body. */
735 /* FIXME - audit the code for other bugs like this one. */
737 adding the C<SvMAGICAL> check to
739 if (AvARYLEN(av) && SvMAGICAL(AvARYLEN(av))) {
740 MAGIC *mg = mg_find (AvARYLEN(av), PERL_MAGIC_arylen);
742 Go through the core and look for similar assumptions that SVs have particular
743 types, as all bets are off during global destruction.
745 =head2 Extend PerlIO and PerlIO::Scalar
747 PerlIO::Scalar doesn't know how to truncate(). Implementing this
748 would require extending the PerlIO vtable.
750 Similarly the PerlIO vtable doesn't know about formats (write()), or
751 about stat(), or chmod()/chown(), utime(), or flock().
753 (For PerlIO::Scalar it's hard to see what e.g. mode bits or ownership
756 PerlIO doesn't do directories or symlinks, either: mkdir(), rmdir(),
757 opendir(), closedir(), seekdir(), rewinddir(), glob(); symlink(),
760 See also L</"Virtualize operating system access">.
762 =head2 -C on the #! line
764 It should be possible to make -C work correctly if found on the #! line,
765 given that all perl command line options are strict ASCII, and -C changes
766 only the interpretation of non-ASCII characters, and not for the script file
767 handle. To make it work needs some investigation of the ordering of function
768 calls during startup, and (by implication) a bit of tweaking of that order.
770 =head2 Organize error messages
772 Perl's diagnostics (error messages, see L<perldiag>) could use
773 reorganizing and formalizing so that each error message has its
774 stable-for-all-eternity unique id, categorized by severity, type, and
775 subsystem. (The error messages would be listed in a datafile outside
776 of the Perl source code, and the source code would only refer to the
777 messages by the id.) This clean-up and regularizing should apply
778 for all croak() messages.
780 This would enable all sorts of things: easier translation/localization
781 of the messages (though please do keep in mind the caveats of
782 L<Locale::Maketext> about too straightforward approaches to
783 translation), filtering by severity, and instead of grepping for a
784 particular error message one could look for a stable error id. (Of
785 course, changing the error messages by default would break all the
786 existing software depending on some particular error message...)
788 This kind of functionality is known as I<message catalogs>. Look for
789 inspiration for example in the catgets() system, possibly even use it
790 if available-- but B<only> if available, all platforms will B<not>
793 For the really pure at heart, consider extending this item to cover
794 also the warning messages (see L<perllexwarn>, C<warnings.pl>).
796 =head1 Tasks that need a knowledge of the interpreter
798 These tasks would need C knowledge, and knowledge of how the interpreter works,
799 or a willingness to learn.
801 =head2 truncate() prototype
803 The prototype of truncate() is currently C<$$>. It should probably
804 be C<*$> instead. (This is changed in F<opcode.pl>)
806 =head2 decapsulation of smart match argument
808 Currently C<$foo ~~ $object> will die with the message "Smart matching a
809 non-overloaded object breaks encapsulation". It would be nice to allow
810 to bypass this by using explictly the syntax C<$foo ~~ %$object> or
813 =head2 error reporting of [$a ; $b]
815 Using C<;> inside brackets is a syntax error, and we don't propose to change
816 that by giving it any meaning. However, it's not reported very helpfully:
818 $ perl -e '$a = [$b; $c];'
819 syntax error at -e line 1, near "$b;"
820 syntax error at -e line 1, near "$c]"
821 Execution of -e aborted due to compilation errors.
823 It should be possible to hook into the tokeniser or the lexer, so that when a
824 C<;> is parsed where it is not legal as a statement terminator (ie inside
825 C<{}> used as a hashref, C<[]> or C<()>) it issues an error something like
826 I<';' isn't legal inside an expression - if you need multiple statements use a
827 do {...} block>. See the thread starting at
828 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-09/msg00573.html
830 =head2 lexicals used only once
834 $ perl -we '$pie = 42'
835 Name "main::pie" used only once: possible typo at -e line 1.
839 $ perl -we 'my $pie = 42'
841 Logically all lexicals used only once should warn, if the user asks for
842 warnings. An unworked RT ticket (#5087) has been open for almost seven
843 years for this discrepancy.
847 The handling of Unicode is unclean in many places. For example, the regexp
848 engine matches in Unicode semantics whenever the string or the pattern is
849 flagged as UTF-8, but that should not be dependent on an internal storage
850 detail of the string. Likewise, case folding behaviour is dependent on the
851 UTF8 internal flag being on or off.
853 =head2 Properly Unicode safe tokeniser and pads.
855 The tokeniser isn't actually very UTF-8 clean. C<use utf8;> is a hack -
856 variable names are stored in stashes as raw bytes, without the utf-8 flag
857 set. The pad API only takes a C<char *> pointer, so that's all bytes too. The
858 tokeniser ignores the UTF-8-ness of C<PL_rsfp>, or any SVs returned from
859 source filters. All this could be fixed.
861 =head2 state variable initialization in list context
863 Currently this is illegal:
865 state ($a, $b) = foo();
867 In Perl 6, C<state ($a) = foo();> and C<(state $a) = foo();> have different
868 semantics, which is tricky to implement in Perl 5 as currently they produce
869 the same opcode trees. The Perl 6 design is firm, so it would be good to
870 implement the necessary code in Perl 5. There are comments in
871 C<Perl_newASSIGNOP()> that show the code paths taken by various assignment
872 constructions involving state variables.
874 =head2 Implement $value ~~ 0 .. $range
876 It would be nice to extend the syntax of the C<~~> operator to also
877 understand numeric (and maybe alphanumeric) ranges.
879 =head2 A does() built-in
881 Like ref(), only useful. It would call the C<DOES> method on objects; it
882 would also tell whether something can be dereferenced as an
883 array/hash/etc., or used as a regexp, etc.
884 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-03/msg00481.html>
886 =head2 Tied filehandles and write() don't mix
888 There is no method on tied filehandles to allow them to be called back by
891 =head2 Propagate compilation hints to the debugger
893 Currently a debugger started with -dE on the command-line doesn't see the
894 features enabled by -E. More generally hints (C<$^H> and C<%^H>) aren't
895 propagated to the debugger. Probably it would be a good thing to propagate
896 hints from the innermost non-C<DB::> scope: this would make code eval'ed
897 in the debugger see the features (and strictures, etc.) currently in
900 =head2 Attach/detach debugger from running program
902 The old perltodo notes "With C<gdb>, you can attach the debugger to a running
903 program if you pass the process ID. It would be good to do this with the Perl
904 debugger on a running Perl program, although I'm not sure how it would be
905 done." ssh and screen do this with named pipes in /tmp. Maybe we can too.
907 =head2 LVALUE functions for lists
909 The old perltodo notes that lvalue functions don't work for list or hash
910 slices. This would be good to fix.
912 =head2 regexp optimiser optional
914 The regexp optimiser is not optional. It should configurable to be, to allow
915 its performance to be measured, and its bugs to be easily demonstrated.
917 =head2 delete &function
919 Allow to delete functions. One can already undef them, but they're still
922 =head2 C</w> regex modifier
924 That flag would enable to match whole words, and also to interpolate
925 arrays as alternations. With it, C</P/w> would be roughly equivalent to:
927 do { local $"='|'; /\b(?:P)\b/ }
929 See L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-01/msg00400.html>
932 =head2 optional optimizer
934 Make the peephole optimizer optional. Currently it performs two tasks as
935 it walks the optree - genuine peephole optimisations, and necessary fixups of
936 ops. It would be good to find an efficient way to switch out the
937 optimisations whilst keeping the fixups.
939 =head2 You WANT *how* many
941 Currently contexts are void, scalar and list. split has a special mechanism in
942 place to pass in the number of return values wanted. It would be useful to
943 have a general mechanism for this, backwards compatible and little speed hit.
944 This would allow proposals such as short circuiting sort to be implemented
947 =head2 lexical aliases
949 Allow lexical aliases (maybe via the syntax C<my \$alias = \$foo>.
951 =head2 entersub XS vs Perl
953 At the moment pp_entersub is huge, and has code to deal with entering both
954 perl and XS subroutines. Subroutine implementations rarely change between
955 perl and XS at run time, so investigate using 2 ops to enter subs (one for
956 XS, one for perl) and swap between if a sub is redefined.
960 Self-ties are currently illegal because they caused too many segfaults. Maybe
961 the causes of these could be tracked down and self-ties on all types
964 =head2 Optimize away @_
966 The old perltodo notes "Look at the "reification" code in C<av.c>".
968 =head2 Virtualize operating system access
970 Implement a set of "vtables" that virtualizes operating system access
971 (open(), mkdir(), unlink(), readdir(), getenv(), etc.) At the very
972 least these interfaces should take SVs as "name" arguments instead of
973 bare char pointers; probably the most flexible and extensible way
974 would be for the Perl-facing interfaces to accept HVs. The system
975 needs to be per-operating-system and per-file-system
976 hookable/filterable, preferably both from XS and Perl level
977 (L<perlport/"Files and Filesystems"> is good reading at this point,
978 in fact, all of L<perlport> is.)
980 This has actually already been implemented (but only for Win32),
981 take a look at F<iperlsys.h> and F<win32/perlhost.h>. While all Win32
982 variants go through a set of "vtables" for operating system access,
983 non-Win32 systems currently go straight for the POSIX/UNIX-style
984 system/library call. Similar system as for Win32 should be
985 implemented for all platforms. The existing Win32 implementation
986 probably does not need to survive alongside this proposed new
987 implementation, the approaches could be merged.
989 What would this give us? One often-asked-for feature this would
990 enable is using Unicode for filenames, and other "names" like %ENV,
991 usernames, hostnames, and so forth.
992 (See L<perlunicode/"When Unicode Does Not Happen">.)
994 But this kind of virtualization would also allow for things like
995 virtual filesystems, virtual networks, and "sandboxes" (though as long
996 as dynamic loading of random object code is allowed, not very safe
997 sandboxes since external code of course know not of Perl's vtables).
998 An example of a smaller "sandbox" is that this feature can be used to
999 implement per-thread working directories: Win32 already does this.
1001 See also L</"Extend PerlIO and PerlIO::Scalar">.
1003 =head2 Investigate PADTMP hash pessimisation
1005 The peephole optimiser converts constants used for hash key lookups to shared
1006 hash key scalars. Under ithreads, something is undoing this work.
1007 See http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-09/msg00793.html
1009 =head2 Store the current pad in the OP slab allocator
1012 I hope that I got that "current pad" part correct
1014 Currently we leak ops in various cases of parse failure. I suggested that we
1015 could solve this by always using the op slab allocator, and walking it to
1016 free ops. Dave comments that as some ops are already freed during optree
1017 creation one would have to mark which ops are freed, and not double free them
1018 when walking the slab. He notes that one problem with this is that for some ops
1019 you have to know which pad was current at the time of allocation, which does
1020 change. I suggested storing a pointer to the current pad in the memory allocated
1021 for the slab, and swapping to a new slab each time the pad changes. Dave thinks
1022 that this would work.
1024 =head2 repack the optree
1026 Repacking the optree after execution order is determined could allow
1027 removal of NULL ops, and optimal ordering of OPs with respect to cache-line
1028 filling. The slab allocator could be reused for this purpose. I think that
1029 the best way to do this is to make it an optional step just before the
1030 completed optree is attached to anything else, and to use the slab allocator
1031 unchanged, so that freeing ops is identical whether or not this step runs.
1032 Note that the slab allocator allocates ops downwards in memory, so one would
1033 have to actually "allocate" the ops in reverse-execution order to get them
1034 contiguous in memory in execution order.
1036 See http://www.nntp.perl.org/group/perl.perl5.porters/2007/12/msg131975.html
1038 Note that running this copy, and then freeing all the old location ops would
1039 cause their slabs to be freed, which would eliminate possible memory wastage if
1040 the previous suggestion is implemented, and we swap slabs more frequently.
1042 =head2 eliminate incorrect line numbers in warnings
1050 } elsif ($undef == 0) {
1053 used to produce this output:
1055 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1056 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1058 where the line of the second warning was misreported - it should be line 5.
1059 Rafael fixed this - the problem arose because there was no nextstate OP
1060 between the execution of the C<if> and the C<elsif>, hence C<PL_curcop> still
1061 reports that the currently executing line is line 4. The solution was to inject
1062 a nextstate OPs for each C<elsif>, although it turned out that the nextstate
1063 OP needed to be a nulled OP, rather than a live nextstate OP, else other line
1064 numbers became misreported. (Jenga!)
1066 The problem is more general than C<elsif> (although the C<elsif> case is the
1067 most common and the most confusing). Ideally this code
1077 would produce this output
1079 Use of uninitialized value $undef in addition (+) at wrong.pl line 4.
1080 Use of uninitialized value $undef in addition (+) at wrong.pl line 7.
1082 (rather than lines 4 and 5), but this would seem to require every OP to carry
1083 (at least) line number information.
1085 What might work is to have an optional line number in memory just before the
1086 BASEOP structure, with a flag bit in the op to say whether it's present.
1087 Initially during compile every OP would carry its line number. Then add a late
1088 pass to the optimiser (potentially combined with L</repack the optree>) which
1089 looks at the two ops on every edge of the graph of the execution path. If
1090 the line number changes, flags the destination OP with this information.
1091 Once all paths are traced, replace every op with the flag with a
1092 nextstate-light op (that just updates C<PL_curcop>), which in turn then passes
1093 control on to the true op. All ops would then be replaced by variants that
1094 do not store the line number. (Which, logically, why it would work best in
1095 conjunction with L</repack the optree>, as that is already copying/reallocating
1098 (Although I should note that we're not certain that doing this for the general
1101 =head2 optimize tail-calls
1103 Tail-calls present an opportunity for broadly applicable optimization;
1104 anywhere that C<< return foo(...) >> is called, the outer return can
1105 be replaced by a goto, and foo will return directly to the outer
1106 caller, saving (conservatively) 25% of perl's call&return cost, which
1107 is relatively higher than in C. The scheme language is known to do
1108 this heavily. B::Concise provides good insight into where this
1109 optimization is possible, ie anywhere entersub,leavesub op-sequence
1112 perl -MO=Concise,-exec,a,b,-main -e 'sub a{ 1 }; sub b {a()}; b(2)'
1114 Bottom line on this is probably a new pp_tailcall function which
1115 combines the code in pp_entersub, pp_leavesub. This should probably
1116 be done 1st in XS, and using B::Generate to patch the new OP into the
1121 Tasks that will get your name mentioned in the description of the "Highlights
1124 =head2 make ithreads more robust
1126 Generally make ithreads more robust. See also L</iCOW>
1128 This task is incremental - even a little bit of work on it will help, and
1129 will be greatly appreciated.
1131 One bit would be to write the missing code in sv.c:Perl_dirp_dup.
1133 Fix Perl_sv_dup, et al so that threads can return objects.
1137 Sarathy and Arthur have a proposal for an improved Copy On Write which
1138 specifically will be able to COW new ithreads. If this can be implemented
1139 it would be a good thing.
1141 =head2 (?{...}) closures in regexps
1143 Fix (or rewrite) the implementation of the C</(?{...})/> closures.
1145 =head2 A re-entrant regexp engine
1147 This will allow the use of a regex from inside (?{ }), (??{ }) and
1148 (?(?{ })|) constructs.
1150 =head2 Add class set operations to regexp engine
1152 Apparently these are quite useful. Anyway, Jeffery Friedl wants them.
1154 demerphq has this on his todo list, but right at the bottom.
1157 =head1 Tasks for microperl
1160 [ Each and every one of these may be obsolete, but they were listed
1161 in the old Todo.micro file]
1164 =head2 make creating uconfig.sh automatic
1166 =head2 make creating Makefile.micro automatic
1168 =head2 do away with fork/exec/wait?
1170 (system, popen should be enough?)
1172 =head2 some of the uconfig.sh really needs to be probed (using cc) in buildtime:
1174 (uConfigure? :-) native datatype widths and endianness come to mind