This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
-D_FORTIFY_SOURCE is no more a todo.
[perl5.git] / Porting / todo.pod
CommitLineData
7711098a
GS
1=head1 NAME
2
c3143508 3todo - Perl TO-DO list
7711098a
GS
4
5=head1 DESCRIPTION
e50bb9a1 6
049aabcb 7This is a list of wishes for Perl. The most up to date version of this file
c3143508 8is at L<http://perl5.git.perl.org/perl.git/blob_plain/HEAD:/Porting/todo.pod>
049aabcb
NC
9
10The tasks we think are smaller or easier are listed first. Anyone is welcome
11to work on any of these, but it's a good idea to first contact
12I<perl5-porters@perl.org> to avoid duplication of effort, and to learn from
13any previous attempts. By all means contact a pumpking privately first if you
14prefer.
e50bb9a1 15
0bdfc961
NC
16Whilst patches to make the list shorter are most welcome, ideas to add to
17the list are also encouraged. Check the perl5-porters archives for past
b4af8972
RB
18ideas, and any discussion about them. One set of archives may be found at
19L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/>
938c8732 20
617eabfa
NC
21What can we offer you in return? Fame, fortune, and everlasting glory? Maybe
22not, but if your patch is incorporated, then we'll add your name to the
23F<AUTHORS> file, which ships in the official distribution. How many other
24programming languages offer you 1 line of immortality?
938c8732 25
0865247f
KW
26=head1 Tasks that need only a little Perl knowledge
27
28=head2 Fix POD errors in Perl documentation
29
30Perl documentation is furnished in POD (Plain Old Documentation); see
31L<perlpod>. We also have a utility that checks for various errors in
32this documentation: F<t/porting/podcheck.t>. Unfortunately many files
33have errors in them, and there is a database of known problems, kept in
34F<t/porting/known_pod_issues.dat>. The most prevalent errors are lines
35too wide to fit in a standard terminal window, but there are more
36serious problems as well; and there are items listed there that are not
37in fact errors. The task would be to go through and clean up the
38documentation. This would be a good way to learn more about Perl.
39
0bdfc961 40=head1 Tasks that only need Perl knowledge
e50bb9a1 41
aa384da9
KW
42=head2 Classify bug tickets by type
43
1e5abc58 44Known bugs in Perl are tracked by L<https://rt.perl.org/> (which also
aa384da9 45includes Perl 6). A summary can be found at
1e5abc58 46L<https://rt.perl.org/NoAuth/perl5/Overview.html>.
aa384da9
KW
47It shows bugs classified by "type". However, the type of many of the
48bugs is "unknown". This greatly lowers the chances of them getting
49fixed, as the number of open bugs is overwhelming -- too many to wade
50through for someone to try to find the bugs in the parts of
51Perl that s/he knows well enough to try to fix. This task involves
52going through these bugs and classifying them into one or more types.
53
54=head2 Ongoing: investigate new bug reports
55
56When a bug report is filed, it would be very helpful to have someone do
57a quick investigation to see if it is a real problem, and to reply to
58the poster about it, asking for example code that reproduces the
59problem. Such code should be added to the test suite as TODO tests, and
60the ticket should be classified by type. To get started on this task,
61look at the tickets that are marked as "New Issues" in
1e5abc58 62L<https://rt.perl.org/NoAuth/perl5/Overview.html>.
aa384da9 63
de2b17d8
NC
64=head2 Migrate t/ from custom TAP generation
65
66Many tests below F<t/> still generate TAP by "hand", rather than using library
96090e4f 67functions. As explained in L<perlhack/TESTING>, tests in F<t/> are
de2b17d8
NC
68written in a particular way to test that more complex constructions actually
69work before using them routinely. Hence they don't use C<Test::More>, but
70instead there is an intentionally simpler library, F<t/test.pl>. However,
71quite a few tests in F<t/> have not been refactored to use it. Refactoring
72any of these tests, one at a time, is a useful thing TODO.
73
0d8e5a42
RGS
74The subdirectories F<base>, F<cmd> and F<comp>, that contain the most
75basic tests, should be excluded from this task.
76
0be987a2
NC
77=head2 Automate perldelta generation
78
79The perldelta file accompanying each release summaries the major changes.
80It's mostly manually generated currently, but some of that could be
81automated with a bit of perl, specifically the generation of
82
83=over
84
85=item Modules and Pragmata
86
87=item New Documentation
88
89=item New Tests
90
91=back
92
93See F<Porting/how_to_write_a_perldelta.pod> for details.
94
0bdfc961 95=head2 Make Schwern poorer
e50bb9a1 96
613bd4f7 97We should have tests for everything. When all the core's modules are tested,
0bdfc961
NC
98Schwern has promised to donate to $500 to TPF. We may need volunteers to
99hold him upside down and shake vigorously in order to actually extract the
100cash.
3958b146 101
831efc5a
JK
102=head2 Write descriptions for all tests
103
104Many individual tests in the test suite lack descriptions (or names, or labels
105-- call them what you will). Many files completely lack descriptions, meaning
106that the only output you get is the test numbers. If all tests had
107descriptions, understanding what the tests are testing and why they sometimes
108fail would both get a whole lot easier.
109
0bdfc961 110=head2 Improve the coverage of the core tests
e50bb9a1 111
e1020413 112Use Devel::Cover to ascertain the core modules' test coverage, then add
02f21748 113tests that are currently missing.
30222c0f 114
0bdfc961 115=head2 test B
e50bb9a1 116
0bdfc961 117A full test suite for the B module would be nice.
e50bb9a1 118
0bdfc961 119=head2 A decent benchmark
e50bb9a1 120
617eabfa 121C<perlbench> seems impervious to any recent changes made to the perl core. It
0bdfc961
NC
122would be useful to have a reasonable general benchmarking suite that roughly
123represented what current perl programs do, and measurably reported whether
124tweaks to the core improve, degrade or don't really affect performance, to
125guide people attempting to optimise the guts of perl. Gisle would welcome
4e1c9055
NC
126new tests for perlbench. Steffen Schwingon would welcome help with
127L<Benchmark::Perl::Formance>
6168cf99 128
0bdfc961 129=head2 fix tainting bugs
6168cf99 130
82046965
FC
131Fix the bugs revealed by running the test suite with the C<-t> switch.
132Setting the TEST_ARGS environment variable to C<-taintwarn> will accomplish
133this.
e50bb9a1 134
0bdfc961 135=head2 Dual life everything
e50bb9a1 136
0bdfc961
NC
137As part of the "dists" plan, anything that doesn't belong in the smallest perl
138distribution needs to be dual lifed. Anything else can be too. Figure out what
139changes would be needed to package that module and its tests up for CPAN, and
140do so. Test it with older perl releases, and fix the problems you find.
e50bb9a1 141
a393eb28
RGS
142To make a minimal perl distribution, it's useful to look at
143F<t/lib/commonsense.t>.
144
0bdfc961 145=head2 POSIX memory footprint
e50bb9a1 146
0bdfc961
NC
147Ilya observed that use POSIX; eats memory like there's no tomorrow, and at
148various times worked to cut it down. There is probably still fat to cut out -
149for example POSIX passes Exporter some very memory hungry data structures.
e50bb9a1 150
8c422da5
NC
151=head2 makedef.pl and conditional compilation
152
153The script F<makedef.pl> that generates the list of exported symbols on
154platforms which need this. Functions are declared in F<embed.fnc>, variables
155in F<intrpvar.h>. Quite a few of the functions and variables are conditionally
156declared there, using C<#ifdef>. However, F<makedef.pl> doesn't understand the
157C macros, so the rules about which symbols are present when is duplicated in
158the Perl code. Writing things twice is bad, m'kay. It would be good to teach
159F<.pl> to understand the conditional compilation, and hence remove the
160duplication, and the mistakes it has caused.
e50bb9a1 161
801de10e
NC
162=head2 use strict; and AutoLoad
163
164Currently if you write
165
166 package Whack;
167 use AutoLoader 'AUTOLOAD';
168 use strict;
169 1;
170 __END__
171 sub bloop {
172 print join (' ', No, strict, here), "!\n";
173 }
174
175then C<use strict;> isn't in force within the autoloaded subroutines. It would
176be more consistent (and less surprising) to arrange for all lexical pragmas
177in force at the __END__ block to be in force within each autoloaded subroutine.
178
773b3597
RGS
179There's a similar problem with SelfLoader.
180
91d0cbf6
NC
181=head2 profile installman
182
183The F<installman> script is slow. All it is doing text processing, which we're
184told is something Perl is good at. So it would be nice to know what it is doing
185that is taking so much CPU, and where possible address it.
186
c69ca1d4 187=head2 enable lexical enabling/disabling of individual warnings
a9ed9b74
JV
188
189Currently, warnings can only be enabled or disabled by category. There
190are times when it would be useful to quash a single warning, not a
191whole category.
91d0cbf6 192
85234543
KW
193=head2 document diagnostics
194
195Many diagnostic messages are not currently documented. The list is at the end
196of t/porting/diag.t.
197
aa384da9
KW
198=head2 Write TODO tests for open bugs
199
200Sometimes bugs get fixed as a side effect of something else, and
201the bug remains open because no one realizes that it has been fixed.
202Ideally, every open bug should have a TODO test in the core test suite.
203
0bdfc961 204=head1 Tasks that need a little sysadmin-type knowledge
e50bb9a1 205
0bdfc961
NC
206Or if you prefer, tasks that you would learn from, and broaden your skills
207base...
e50bb9a1 208
cd793d32 209=head2 make HTML install work
e50bb9a1 210
78b489b0 211There is an C<install.html> target in the Makefile. It's marked as
adebf063
NC
212"experimental". It would be good to get this tested, make it work reliably, and
213remove the "experimental" tag. This would include
214
215=over 4
216
217=item 1
218
219Checking that cross linking between various parts of the documentation works.
220In particular that links work between the modules (files with POD in F<lib/>)
221and the core documentation (files in F<pod/>)
222
223=item 2
224
78b489b0
NC
225Improving the code that split C<perlfunc> into chunks, preferably with
226general case code added to L<Pod::Functions> that could be used elsewhere.
227
617eabfa
NC
228Challenges here are correctly identifying the groups of functions that go
229together, and making the right named external cross-links point to the right
78b489b0
NC
230page. Currently this works reasonably well in the general case, and correctly
231parses two or more C<=items> giving the different parameter lists for the
232same function, such used by C<substr>. However it fails completely where
233I<different> functions are listed as a sequence of C<=items> but share the
234same description. All the functions from C<getpwnam> to C<endprotoent> have
235individual stub pages, with only the page for C<endservent> holding the
236description common to all. Likewise C<q>, C<qq> and C<qw> have stub pages,
237instead of sharing the body of C<qx>.
238
239Note also the current code isn't ideal with the two forms of C<select>, mushing
240them both into one F<select.html> with the two descriptions run together.
241Fixing this may well be a special case.
adebf063
NC
242
243=back
3a89a73c 244
0bdfc961
NC
245=head2 compressed man pages
246
247Be able to install them. This would probably need a configure test to see how
248the system does compressed man pages (same directory/different directory?
249same filename/different filename), as well as tweaking the F<installman> script
250to compress as necessary.
251
30222c0f
NC
252=head2 Add a code coverage target to the Makefile
253
254Make it easy for anyone to run Devel::Cover on the core's tests. The steps
255to do this manually are roughly
256
257=over 4
258
259=item *
260
261do a normal C<Configure>, but include Devel::Cover as a module to install
f11a3063 262(see L<INSTALL> for how to do this)
30222c0f
NC
263
264=item *
265
266 make perl
267
268=item *
269
270 cd t; HARNESS_PERL_SWITCHES=-MDevel::Cover ./perl -I../lib harness
271
272=item *
273
274Process the resulting Devel::Cover database
275
276=back
277
278This just give you the coverage of the F<.pm>s. To also get the C level
279coverage you need to
280
281=over 4
282
283=item *
284
285Additionally tell C<Configure> to use the appropriate C compiler flags for
286C<gcov>
287
288=item *
289
290 make perl.gcov
291
292(instead of C<make perl>)
293
294=item *
295
296After running the tests run C<gcov> to generate all the F<.gcov> files.
297(Including down in the subdirectories of F<ext/>
298
299=item *
300
301(From the top level perl directory) run C<gcov2perl> on all the C<.gcov> files
302to get their stats into the cover_db directory.
303
304=item *
305
306Then process the Devel::Cover database
307
308=back
309
310It would be good to add a single switch to C<Configure> to specify that you
311wanted to perform perl level coverage, and another to specify C level
312coverage, and have C<Configure> and the F<Makefile> do all the right things
313automatically.
314
02f21748 315=head2 Make Config.pm cope with differences between built and installed perl
0bdfc961
NC
316
317Quite often vendors ship a perl binary compiled with their (pay-for)
318compilers. People install a free compiler, such as gcc. To work out how to
319build extensions, Perl interrogates C<%Config>, so in this situation
320C<%Config> describes compilers that aren't there, and extension building
321fails. This forces people into choosing between re-compiling perl themselves
322using the compiler they have, or only using modules that the vendor ships.
323
324It would be good to find a way teach C<Config.pm> about the installation setup,
325possibly involving probing at install time or later, so that the C<%Config> in
326a binary distribution better describes the installed machine, when the
327installed machine differs from the build machine in some significant way.
328
728f4ecd
NC
329=head2 linker specification files
330
331Some platforms mandate that you provide a list of a shared library's external
332symbols to the linker, so the core already has the infrastructure in place to
4e1c9055
NC
333do this for generating shared perl libraries. Florian Ragwitz has been working
334to offer this for the GNU toolchain, to allow Unix users to test that the
728f4ecd 335export list is correct, and to build a perl that does not pollute the global
32d539f5 336namespace with private symbols, and will fail in the same way as msvc or mingw
4e1c9055 337builds or when using PERL_DL_NONLAZY=1. See the branch smoke-me/rafl/ld_export
728f4ecd 338
a229ae3b
RGS
339=head2 Cross-compile support
340
4e1c9055
NC
341We get requests for "how to cross compile Perl". The vast majority of these
342seem to be for a couple of scenarios:
343
344=over 4
345
346=item *
347
348Platforms that could build natively using F<./Configure> (I<e.g.> Linux or
349NetBSD on MIPS or ARM) but people want to use a beefier machine (and on the
350same OS) to build more easily.
351
352=item *
353
354Platforms that can't build natively, but no (significant) porting changes
355are needed to our current source code. Prime example of this is Android.
356
357=back
358
359There are several scripts and tools for cross-compiling perl for other
360platforms. However, these are somewhat inconsistent and scattered across the
361codebase, none are documented well, none are clearly flexible enough to
c5fb089a 362be confident that they can support any TARGET/HOST platform pair other than
4e1c9055
NC
363that which they were developed on, and it's not clear how bitrotted they are.
364
365For example, C<Configure> understands C<-Dusecrosscompile> option. This option
a229ae3b 366arranges for building C<miniperl> for TARGET machine, so this C<miniperl> is
4e1c9055
NC
367assumed then to be copied to TARGET machine and used as a replacement of
368full C<perl> executable. This code is almost 10 years old. Meanwhile, the
369F<Cross/> directory contains two different approaches for cross compiling to
370ARM Linux targets, relying on hand curated F<config.sh> files, but that code
371is getting on for 5 years old, and requires insider knowledge of perl's
372build system to draft a F<config.sh> for a new platform.
373
c5fb089a 374Jess Robinson has submitted a grant to TPF to work on cleaning this up.
0bdfc961 375
98fca0e8
NC
376=head2 Split "linker" from "compiler"
377
378Right now, Configure probes for two commands, and sets two variables:
379
380=over 4
381
b91dd380 382=item * C<cc> (in F<cc.U>)
98fca0e8
NC
383
384This variable holds the name of a command to execute a C compiler which
385can resolve multiple global references that happen to have the same
386name. Usual values are F<cc> and F<gcc>.
387Fervent ANSI compilers may be called F<c89>. AIX has F<xlc>.
388
b91dd380 389=item * C<ld> (in F<dlsrc.U>)
98fca0e8
NC
390
391This variable indicates the program to be used to link
392libraries for dynamic loading. On some systems, it is F<ld>.
393On ELF systems, it should be C<$cc>. Mostly, we'll try to respect
394the hint file setting.
395
396=back
397
8d159ec1
NC
398There is an implicit historical assumption from around Perl5.000alpha
399something, that C<$cc> is also the correct command for linking object files
400together to make an executable. This may be true on Unix, but it's not true
401on other platforms, and there are a maze of work arounds in other places (such
402as F<Makefile.SH>) to cope with this.
98fca0e8
NC
403
404Ideally, we should create a new variable to hold the name of the executable
405linker program, probe for it in F<Configure>, and centralise all the special
406case logic there or in hints files.
407
408A small bikeshed issue remains - what to call it, given that C<$ld> is already
8d159ec1
NC
409taken (arguably for the wrong thing now, but on SunOS 4.1 it is the command
410for creating dynamically-loadable modules) and C<$link> could be confused with
411the Unix command line executable of the same name, which does something
412completely different. Andy Dougherty makes the counter argument "In parrot, I
413tried to call the command used to link object files and libraries into an
414executable F<link>, since that's what my vaguely-remembered DOS and VMS
415experience suggested. I don't think any real confusion has ensued, so it's
416probably a reasonable name for perl5 to use."
98fca0e8
NC
417
418"Alas, I've always worried that introducing it would make things worse,
419since now the module building utilities would have to look for
420C<$Config{link}> and institute a fall-back plan if it weren't found."
8d159ec1
NC
421Although I can see that as confusing, given that C<$Config{d_link}> is true
422when (hard) links are available.
98fca0e8 423
75585ce3
SP
424=head2 Configure Windows using PowerShell
425
426Currently, Windows uses hard-coded config files based to build the
427config.h for compiling Perl. Makefiles are also hard-coded and need to be
428hand edited prior to building Perl. While this makes it easy to create a perl.exe
429that works across multiple Windows versions, being able to accurately
430configure a perl.exe for a specific Windows versions and VS C++ would be
431a nice enhancement. With PowerShell available on Windows XP and up, this
432may now be possible. Step 1 might be to investigate whether this is possible
433and use this to clean up our current makefile situation. Step 2 would be to
434see if there would be a way to use our existing metaconfig units to configure a
435Windows Perl or whether we go in a separate direction and make it so. Of
436course, we all know what step 3 is.
437
0bdfc961
NC
438=head1 Tasks that need a little C knowledge
439
440These tasks would need a little C knowledge, but don't need any specific
441background or experience with XS, or how the Perl interpreter works
442
3d826b29
NC
443=head2 Weed out needless PERL_UNUSED_ARG
444
445The C code uses the macro C<PERL_UNUSED_ARG> to stop compilers warning about
446unused arguments. Often the arguments can't be removed, as there is an
447external constraint that determines the prototype of the function, so this
448approach is valid. However, there are some cases where C<PERL_UNUSED_ARG>
449could be removed. Specifically
450
451=over 4
452
453=item *
454
455The prototypes of (nearly all) static functions can be changed
456
457=item *
458
459Unused arguments generated by short cut macros are wasteful - the short cut
460macro used can be changed.
461
462=back
463
bcbaa2d5
RGS
464=head2 -Duse32bit*
465
466Natively 64-bit systems need neither -Duse64bitint nor -Duse64bitall.
467On these systems, it might be the default compilation mode, and there
468is currently no guarantee that passing no use64bitall option to the
469Configure process will build a 32bit perl. Implementing -Duse32bit*
f6ca94fb 470options would be nice for perl 5.21.2.
bcbaa2d5 471
fee0a0f7 472=head2 Profile Perl - am I hot or not?
62403a3c 473
fee0a0f7
NC
474The Perl source code is stable enough that it makes sense to profile it,
475identify and optimise the hotspots. It would be good to measure the
476performance of the Perl interpreter using free tools such as cachegrind,
477gprof, and dtrace, and work to reduce the bottlenecks they reveal.
478
479As part of this, the idea of F<pp_hot.c> is that it contains the I<hot> ops,
480the ops that are most commonly used. The idea is that by grouping them, their
481object code will be adjacent in the executable, so they have a greater chance
482of already being in the CPU cache (or swapped in) due to being near another op
483already in use.
62403a3c
NC
484
485Except that it's not clear if these really are the most commonly used ops. So
fee0a0f7
NC
486as part of exercising your skills with coverage and profiling tools you might
487want to determine what ops I<really> are the most commonly used. And in turn
488suggest evictions and promotions to achieve a better F<pp_hot.c>.
62403a3c 489
91d0cbf6
NC
490One piece of Perl code that might make a good testbed is F<installman>.
491
a229ae3b 492=head2 Improve win32/wince.c
0bdfc961 493
a229ae3b 494Currently, numerous functions look virtually, if not completely,
c23989d1 495identical in both F<win32/wince.c> and F<win32/win32.c> files, which can't
6d71adcd
NC
496be good.
497
c5b31784
SH
498=head2 Use secure CRT functions when building with VC8 on Win32
499
500Visual C++ 2005 (VC++ 8.x) deprecated a number of CRT functions on the basis
501that they were "unsafe" and introduced differently named secure versions of
502them as replacements, e.g. instead of writing
503
504 FILE* f = fopen(__FILE__, "r");
505
506one should now write
507
508 FILE* f;
509 errno_t err = fopen_s(&f, __FILE__, "r");
510
511Currently, the warnings about these deprecations have been disabled by adding
512-D_CRT_SECURE_NO_DEPRECATE to the CFLAGS. It would be nice to remove that
513warning suppressant and actually make use of the new secure CRT functions.
514
515There is also a similar issue with POSIX CRT function names like fileno having
516been deprecated in favour of ISO C++ conformant names like _fileno. These
26a6faa8 517warnings are also currently suppressed by adding -D_CRT_NONSTDC_NO_DEPRECATE. It
c5b31784
SH
518might be nice to do as Microsoft suggest here too, although, unlike the secure
519functions issue, there is presumably little or no benefit in this case.
520
038ae9a4
SH
521=head2 Fix POSIX::access() and chdir() on Win32
522
523These functions currently take no account of DACLs and therefore do not behave
524correctly in situations where access is restricted by DACLs (as opposed to the
525read-only attribute).
526
527Furthermore, POSIX::access() behaves differently for directories having the
528read-only attribute set depending on what CRT library is being used. For
529example, the _access() function in the VC6 and VC7 CRTs (wrongly) claim that
530such directories are not writable, whereas in fact all directories are writable
531unless access is denied by DACLs. (In the case of directories, the read-only
532attribute actually only means that the directory cannot be deleted.) This CRT
533bug is fixed in the VC8 and VC9 CRTs (but, of course, the directory may still
534not actually be writable if access is indeed denied by DACLs).
535
536For the chdir() issue, see ActiveState bug #74552:
b4af8972 537L<http://bugs.activestate.com/show_bug.cgi?id=74552>
038ae9a4
SH
538
539Therefore, DACLs should be checked both for consistency across CRTs and for
540the correct answer.
541
542(Note that perl's -w operator should not be modified to check DACLs. It has
543been written so that it reflects the state of the read-only attribute, even
544for directories (whatever CRT is being used), for symmetry with chmod().)
545
16815324
NC
546=head2 strcat(), strcpy(), strncat(), strncpy(), sprintf(), vsprintf()
547
548Maybe create a utility that checks after each libperl.a creation that
549none of the above (nor sprintf(), vsprintf(), or *SHUDDER* gets())
550ever creep back to libperl.a.
551
1dcc3c19
DG
552 nm libperl.a | ./miniperl -alne '$o = $F[0] if /:$/;
553 print "$o $F[1]" if $F[0] eq "U" && $F[1] =~ /^(?:strn?c(?:at|py)|v?sprintf|gets)$/'
16815324
NC
554
555Note, of course, that this will only tell whether B<your> platform
556is using those naughty interfaces.
557
8964cfe0
NC
558=head2 Arenas for GPs? For MAGIC?
559
560C<struct gp> and C<struct magic> are both currently allocated by C<malloc>.
561It might be a speed or memory saving to change to using arenas. Or it might
562not. It would need some suitable benchmarking first. In particular, C<GP>s
563can probably be changed with minimal compatibility impact (probably nothing
564outside of the core, or even outside of F<gv.c> allocates them), but they
565probably aren't allocated/deallocated often enough for a speed saving. Whereas
566C<MAGIC> is allocated/deallocated more often, but in turn, is also something
567more externally visible, so changing the rules here may bite external code.
568
3880c8ec
NC
569=head2 Shared arenas
570
571Several SV body structs are now the same size, notably PVMG and PVGV, PVAV and
572PVHV, and PVCV and PVFM. It should be possible to allocate and return same
573sized bodies from the same actual arena, rather than maintaining one arena for
574each. This could save 4-6K per thread, of memory no longer tied up in the
575not-yet-allocated part of an arena.
576
8964cfe0 577
6d71adcd
NC
578=head1 Tasks that need a knowledge of XS
579
580These tasks would need C knowledge, and roughly the level of knowledge of
581the perl API that comes from writing modules that use XS to interface to
582C.
583
e851c105
DG
584=head2 Write an XS cookbook
585
586Create pod/perlxscookbook.pod with short, task-focused 'recipes' in XS that
587demonstrate common tasks and good practices. (Some of these might be
588extracted from perlguts.) The target audience should be XS novices, who need
589more examples than perlguts but something less overwhelming than perlapi.
590Recipes should provide "one pretty good way to do it" instead of TIMTOWTDI.
591
5b7d14ff
DG
592Rather than focusing on interfacing Perl to C libraries, such a cookbook
593should probably focus on how to optimize Perl routines by re-writing them
594in XS. This will likely be more motivating to those who mostly work in
595Perl but are looking to take the next step into XS.
596
597Deconstructing and explaining some simpler XS modules could be one way to
598bootstrap a cookbook. (List::Util? Class::XSAccessor? Tree::Ternary_XS?)
599Another option could be deconstructing the implementation of some simpler
600functions in op.c.
601
0b162fb0 602=head2 Document how XSUBs can use C<cv_set_call_checker> to inline themselves as OPs
05fb4e20
NC
603
604For a simple XSUB, often the subroutine dispatch takes more time than the
0b162fb0
NC
605XSUB itself. v5.14.0 now allows XSUBs to register a function which will be
606called when the parser is finished building an C<entersub> op which calls
607them.
608
609Registration is done with C<Perl_cv_set_call_checker>, is documented at the
610API level in L<perlapi>, and L<perl5140delta/Custom per-subroutine check hooks>
611notes that it can be used to inline a subroutine, by replacing it with a
612custom op. However there is no further detail of the code needed to do this.
613It would be useful to add one or more annotated examples of how to create
614XSUBs that inline.
615
616This should provide a measurable speed up to simple XSUBs inside
05fb4e20
NC
617tight loops. Initially one would have to write the OP alternative
618implementation by hand, but it's likely that this should be reasonably
619straightforward for the type of XSUB that would benefit the most. Longer
620term, once the run-time implementation is proven, it should be possible to
621progressively update ExtUtils::ParseXS to generate OP implementations for
622some XSUBs.
623
318bf708
NC
624=head2 Remove the use of SVs as temporaries in dump.c
625
626F<dump.c> contains debugging routines to dump out the contains of perl data
627structures, such as C<SV>s, C<AV>s and C<HV>s. Currently, the dumping code
628B<uses> C<SV>s for its temporary buffers, which was a logical initial
629implementation choice, as they provide ready made memory handling.
630
631However, they also lead to a lot of confusion when it happens that what you're
632trying to debug is seen by the code in F<dump.c>, correctly or incorrectly, as
633a temporary scalar it can use for a temporary buffer. It's also not possible
634to dump scalars before the interpreter is properly set up, such as during
635ithreads cloning. It would be good to progressively replace the use of scalars
636as string accumulation buffers with something much simpler, directly allocated
637by C<malloc>. The F<dump.c> code is (or should be) only producing 7 bit
638US-ASCII, so output character sets are not an issue.
639
640Producing and proving an internal simple buffer allocation would make it easier
641to re-write the internals of the PerlIO subsystem to avoid using C<SV>s for
642B<its> buffers, use of which can cause problems similar to those of F<dump.c>,
643at similar times.
644
5d96f598
NC
645=head2 safely supporting POSIX SA_SIGINFO
646
647Some years ago Jarkko supplied patches to provide support for the POSIX
648SA_SIGINFO feature in Perl, passing the extra data to the Perl signal handler.
649
650Unfortunately, it only works with "unsafe" signals, because under safe
651signals, by the time Perl gets to run the signal handler, the extra
652information has been lost. Moreover, it's not easy to store it somewhere,
653as you can't call mutexs, or do anything else fancy, from inside a signal
654handler.
655
656So it strikes me that we could provide safe SA_SIGINFO support
657
658=over 4
659
660=item 1
661
662Provide global variables for two file descriptors
663
664=item 2
665
666When the first request is made via C<sigaction> for C<SA_SIGINFO>, create a
667pipe, store the reader in one, the writer in the other
668
669=item 3
670
671In the "safe" signal handler (C<Perl_csighandler()>/C<S_raise_signal()>), if
672the C<siginfo_t> pointer non-C<NULL>, and the writer file handle is open,
673
674=over 8
675
676=item 1
677
678serialise signal number, C<struct siginfo_t> (or at least the parts we care
679about) into a small auto char buff
680
681=item 2
682
683C<write()> that (non-blocking) to the writer fd
684
685=over 12
686
687=item 1
688
689if it writes 100%, flag the signal in a counter of "signals on the pipe" akin
690to the current per-signal-number counts
691
692=item 2
693
694if it writes 0%, assume the pipe is full. Flag the data as lost?
695
696=item 3
697
698if it writes partially, croak a panic, as your OS is broken.
699
700=back
701
702=back
703
704=item 4
705
706in the regular C<PERL_ASYNC_CHECK()> processing, if there are "signals on
707the pipe", read the data out, deserialise, build the Perl structures on
708the stack (code in C<Perl_sighandler()>, the "unsafe" handler), and call as
709usual.
710
711=back
712
713I think that this gets us decent C<SA_SIGINFO> support, without the current risk
714of running Perl code inside the signal handler context. (With all the dangers
715of things like C<malloc> corruption that that currently offers us)
716
717For more information see the thread starting with this message:
b4af8972 718L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-03/msg00305.html>
5d96f598 719
6d71adcd
NC
720=head2 autovivification
721
722Make all autovivification consistent w.r.t LVALUE/RVALUE and strict/no strict;
723
724This task is incremental - even a little bit of work on it will help.
725
726=head2 Unicode in Filenames
727
728chdir, chmod, chown, chroot, exec, glob, link, lstat, mkdir, open,
729opendir, qx, readdir, readlink, rename, rmdir, stat, symlink, sysopen,
730system, truncate, unlink, utime, -X. All these could potentially accept
731Unicode filenames either as input or output (and in the case of system
732and qx Unicode in general, as input or output to/from the shell).
733Whether a filesystem - an operating system pair understands Unicode in
734filenames varies.
735
736Known combinations that have some level of understanding include
737Microsoft NTFS, Apple HFS+ (In Mac OS 9 and X) and Apple UFS (in Mac
738OS X), NFS v4 is rumored to be Unicode, and of course Plan 9. How to
739create Unicode filenames, what forms of Unicode are accepted and used
740(UCS-2, UTF-16, UTF-8), what (if any) is the normalization form used,
741and so on, varies. Finding the right level of interfacing to Perl
742requires some thought. Remember that an OS does not implicate a
743filesystem.
744
745(The Windows -C command flag "wide API support" has been at least
746temporarily retired in 5.8.1, and the -C has been repurposed, see
747L<perlrun>.)
748
87a942b1
JH
749Most probably the right way to do this would be this:
750L</"Virtualize operating system access">.
751
6d71adcd
NC
752=head2 Unicode in %ENV
753
754Currently the %ENV entries are always byte strings.
87a942b1 755See L</"Virtualize operating system access">.
6d71adcd 756
799c141b
SH
757(See RT ticket #113536 for information on Win32's handling of %ENV,
758which was fixed to work with native ANSI codepage characters in the
759environment, but still doesn't work with other characters outside of
760that codepage present in the environment.)
761
1f2e7916
JD
762=head2 Unicode and glob()
763
764Currently glob patterns and filenames returned from File::Glob::glob()
87a942b1 765are always byte strings. See L</"Virtualize operating system access">.
1f2e7916 766
6d71adcd
NC
767=head2 use less 'memory'
768
769Investigate trade offs to switch out perl's choices on memory usage.
770Particularly perl should be able to give memory back.
771
772This task is incremental - even a little bit of work on it will help.
773
774=head2 Re-implement C<:unique> in a way that is actually thread-safe
775
776The old implementation made bad assumptions on several levels. A good 90%
777solution might be just to make C<:unique> work to share the string buffer
778of SvPVs. That way large constant strings can be shared between ithreads,
779such as the configuration information in F<Config>.
780
781=head2 Make tainting consistent
782
783Tainting would be easier to use if it didn't take documented shortcuts and
784allow taint to "leak" everywhere within an expression.
785
786=head2 readpipe(LIST)
787
788system() accepts a LIST syntax (and a PROGRAM LIST syntax) to avoid
789running a shell. readpipe() (the function behind qx//) could be similarly
790extended.
791
6d71adcd
NC
792=head2 Audit the code for destruction ordering assumptions
793
794Change 25773 notes
795
796 /* Need to check SvMAGICAL, as during global destruction it may be that
797 AvARYLEN(av) has been freed before av, and hence the SvANY() pointer
798 is now part of the linked list of SV heads, rather than pointing to
799 the original body. */
800 /* FIXME - audit the code for other bugs like this one. */
801
802adding the C<SvMAGICAL> check to
803
804 if (AvARYLEN(av) && SvMAGICAL(AvARYLEN(av))) {
805 MAGIC *mg = mg_find (AvARYLEN(av), PERL_MAGIC_arylen);
806
807Go through the core and look for similar assumptions that SVs have particular
808types, as all bets are off during global destruction.
809
749904bf
JH
810=head2 Extend PerlIO and PerlIO::Scalar
811
812PerlIO::Scalar doesn't know how to truncate(). Implementing this
813would require extending the PerlIO vtable.
814
815Similarly the PerlIO vtable doesn't know about formats (write()), or
816about stat(), or chmod()/chown(), utime(), or flock().
817
818(For PerlIO::Scalar it's hard to see what e.g. mode bits or ownership
819would mean.)
820
821PerlIO doesn't do directories or symlinks, either: mkdir(), rmdir(),
822opendir(), closedir(), seekdir(), rewinddir(), glob(); symlink(),
823readlink().
824
94da6c29
JH
825See also L</"Virtualize operating system access">.
826
d6c1e11f
JH
827=head2 Organize error messages
828
829Perl's diagnostics (error messages, see L<perldiag>) could use
a8d0aeb9 830reorganizing and formalizing so that each error message has its
d6c1e11f
JH
831stable-for-all-eternity unique id, categorized by severity, type, and
832subsystem. (The error messages would be listed in a datafile outside
c4bd451b
CB
833of the Perl source code, and the source code would only refer to the
834messages by the id.) This clean-up and regularizing should apply
d6c1e11f
JH
835for all croak() messages.
836
837This would enable all sorts of things: easier translation/localization
838of the messages (though please do keep in mind the caveats of
839L<Locale::Maketext> about too straightforward approaches to
840translation), filtering by severity, and instead of grepping for a
841particular error message one could look for a stable error id. (Of
842course, changing the error messages by default would break all the
843existing software depending on some particular error message...)
844
845This kind of functionality is known as I<message catalogs>. Look for
846inspiration for example in the catgets() system, possibly even use it
847if available-- but B<only> if available, all platforms will B<not>
de96509d 848have catgets().
d6c1e11f
JH
849
850For the really pure at heart, consider extending this item to cover
47a9c258 851also the warning messages (see L<warnings>, F<regen/warnings.pl>).
3236f110 852
0bdfc961 853=head1 Tasks that need a knowledge of the interpreter
3298bd4d 854
0bdfc961
NC
855These tasks would need C knowledge, and knowledge of how the interpreter works,
856or a willingness to learn.
3298bd4d 857
10517af5
JD
858=head2 forbid labels with keyword names
859
860Currently C<goto keyword> "computes" the label value:
861
862 $ perl -e 'goto print'
863 Can't find label 1 at -e line 1.
864
343c8006
JD
865It is controversial if the right way to avoid the confusion is to forbid
866labels with keyword names, or if it would be better to always treat
867bareword expressions after a "goto" as a label and never as a keyword.
10517af5 868
de6375e3
RGS
869=head2 truncate() prototype
870
871The prototype of truncate() is currently C<$$>. It should probably
872be C<*$> instead. (This is changed in F<opcode.pl>)
873
565590b5
NC
874=head2 error reporting of [$a ; $b]
875
876Using C<;> inside brackets is a syntax error, and we don't propose to change
877that by giving it any meaning. However, it's not reported very helpfully:
878
879 $ perl -e '$a = [$b; $c];'
880 syntax error at -e line 1, near "$b;"
881 syntax error at -e line 1, near "$c]"
882 Execution of -e aborted due to compilation errors.
883
884It should be possible to hook into the tokeniser or the lexer, so that when a
885C<;> is parsed where it is not legal as a statement terminator (ie inside
886C<{}> used as a hashref, C<[]> or C<()>) it issues an error something like
887I<';' isn't legal inside an expression - if you need multiple statements use a
888do {...} block>. See the thread starting at
b4af8972 889L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-09/msg00573.html>
565590b5 890
e053a921
RS
891=head2 strict as warnings
892
893See L<http://markmail.org/message/vbrupaslr3bybmvk>, where Josua ben Jore
894writes: I've been of the opinion that everything strict.pm does ought to be
895able to considered just warnings that have been promoted to 'FATAL'.
896
718140ec
NC
897=head2 lexicals used only once
898
899This warns:
900
901 $ perl -we '$pie = 42'
902 Name "main::pie" used only once: possible typo at -e line 1.
903
904This does not:
905
906 $ perl -we 'my $pie = 42'
907
908Logically all lexicals used only once should warn, if the user asks for
d6f4ea2e
SP
909warnings. An unworked RT ticket (#5087) has been open for almost seven
910years for this discrepancy.
718140ec 911
a3d15f9a
RGS
912=head2 UTF-8 revamp
913
85c006b6
KW
914The handling of Unicode is unclean in many places. In the regex engine
915there are especially many problems. The swash data structure could be
916replaced my something better. Inversion lists and maps are likely
917candidates. The whole Unicode database could be placed in-core for a
918huge speed-up. Only minimal work was done on the optimizer when utf8
919was added, with the result that the synthetic start class often will
920fail to narrow down the possible choices when given non-Latin1 input.
4e1c9055 921Karl Williamson has been working on this - talk to him.
a3d15f9a 922
636e63cb
NC
923=head2 state variable initialization in list context
924
925Currently this is illegal:
926
927 state ($a, $b) = foo();
928
a2874905 929In Perl 6, C<state ($a) = foo();> and C<(state $a) = foo();> have different
a8d0aeb9 930semantics, which is tricky to implement in Perl 5 as currently they produce
a2874905 931the same opcode trees. The Perl 6 design is firm, so it would be good to
a8d0aeb9 932implement the necessary code in Perl 5. There are comments in
a2874905
NC
933C<Perl_newASSIGNOP()> that show the code paths taken by various assignment
934constructions involving state variables.
636e63cb 935
a393eb28
RGS
936=head2 A does() built-in
937
938Like ref(), only useful. It would call the C<DOES> method on objects; it
939would also tell whether something can be dereferenced as an
940array/hash/etc., or used as a regexp, etc.
941L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-03/msg00481.html>
942
943=head2 Tied filehandles and write() don't mix
944
945There is no method on tied filehandles to allow them to be called back by
946formats.
4fedb12c 947
53967bb9
RGS
948=head2 Propagate compilation hints to the debugger
949
950Currently a debugger started with -dE on the command-line doesn't see the
951features enabled by -E. More generally hints (C<$^H> and C<%^H>) aren't
952propagated to the debugger. Probably it would be a good thing to propagate
953hints from the innermost non-C<DB::> scope: this would make code eval'ed
954in the debugger see the features (and strictures, etc.) currently in
955scope.
956
d10fc472 957=head2 Attach/detach debugger from running program
1626a787 958
cd793d32
NC
959The old perltodo notes "With C<gdb>, you can attach the debugger to a running
960program if you pass the process ID. It would be good to do this with the Perl
0bdfc961
NC
961debugger on a running Perl program, although I'm not sure how it would be
962done." ssh and screen do this with named pipes in /tmp. Maybe we can too.
1626a787 963
0bdfc961
NC
964=head2 LVALUE functions for lists
965
966The old perltodo notes that lvalue functions don't work for list or hash
967slices. This would be good to fix.
968
c5fb089a 969=head2 regexp optimizer optional
0bdfc961 970
c5fb089a
DS
971The regexp optimizer is not optional. It should be configurable to be optional
972and to allow its performance to be measured and its bugs to be easily
973demonstrated.
0bdfc961 974
ef36c6a7
RGS
975=head2 C</w> regex modifier
976
977That flag would enable to match whole words, and also to interpolate
978arrays as alternations. With it, C</P/w> would be roughly equivalent to:
979
980 do { local $"='|'; /\b(?:P)\b/ }
981
b4af8972
RB
982See
983L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-01/msg00400.html>
ef36c6a7
RGS
984for the discussion.
985
0bdfc961
NC
986=head2 optional optimizer
987
988Make the peephole optimizer optional. Currently it performs two tasks as
989it walks the optree - genuine peephole optimisations, and necessary fixups of
990ops. It would be good to find an efficient way to switch out the
991optimisations whilst keeping the fixups.
992
993=head2 You WANT *how* many
994
995Currently contexts are void, scalar and list. split has a special mechanism in
996place to pass in the number of return values wanted. It would be useful to
997have a general mechanism for this, backwards compatible and little speed hit.
998This would allow proposals such as short circuiting sort to be implemented
999as a module on CPAN.
1000
1001=head2 lexical aliases
1002
e12cb30b 1003Allow lexical aliases (maybe via the syntax C<my \$alias = \$foo>).
0bdfc961 1004
de535794 1005=head2 Self-ties
2810d901 1006
de535794 1007Self-ties are currently illegal because they caused too many segfaults. Maybe
a8d0aeb9 1008the causes of these could be tracked down and self-ties on all types
de535794 1009reinstated.
0bdfc961
NC
1010
1011=head2 Optimize away @_
1012
1013The old perltodo notes "Look at the "reification" code in C<av.c>".
1014
87a942b1
JH
1015=head2 Virtualize operating system access
1016
1017Implement a set of "vtables" that virtualizes operating system access
1018(open(), mkdir(), unlink(), readdir(), getenv(), etc.) At the very
1019least these interfaces should take SVs as "name" arguments instead of
1020bare char pointers; probably the most flexible and extensible way
e1a3d5d1
JH
1021would be for the Perl-facing interfaces to accept HVs. The system
1022needs to be per-operating-system and per-file-system
1023hookable/filterable, preferably both from XS and Perl level
87a942b1
JH
1024(L<perlport/"Files and Filesystems"> is good reading at this point,
1025in fact, all of L<perlport> is.)
1026
e1a3d5d1
JH
1027This has actually already been implemented (but only for Win32),
1028take a look at F<iperlsys.h> and F<win32/perlhost.h>. While all Win32
1029variants go through a set of "vtables" for operating system access,
e1020413 1030non-Win32 systems currently go straight for the POSIX/Unix-style
e1a3d5d1
JH
1031system/library call. Similar system as for Win32 should be
1032implemented for all platforms. The existing Win32 implementation
1033probably does not need to survive alongside this proposed new
1034implementation, the approaches could be merged.
87a942b1
JH
1035
1036What would this give us? One often-asked-for feature this would
94da6c29
JH
1037enable is using Unicode for filenames, and other "names" like %ENV,
1038usernames, hostnames, and so forth.
1039(See L<perlunicode/"When Unicode Does Not Happen">.)
1040
1041But this kind of virtualization would also allow for things like
1042virtual filesystems, virtual networks, and "sandboxes" (though as long
1043as dynamic loading of random object code is allowed, not very safe
1044sandboxes since external code of course know not of Perl's vtables).
1045An example of a smaller "sandbox" is that this feature can be used to
1046implement per-thread working directories: Win32 already does this.
1047
1048See also L</"Extend PerlIO and PerlIO::Scalar">.
87a942b1 1049
52960e22
JC
1050=head2 repack the optree
1051
1052Repacking the optree after execution order is determined could allow
057163d7 1053removal of NULL ops, and optimal ordering of OPs with respect to cache-line
2723c0fb 1054filling. I think that
057163d7
NC
1055the best way to do this is to make it an optional step just before the
1056completed optree is attached to anything else, and to use the slab allocator
2723c0fb
FC
1057unchanged--but allocate a single slab the right size, avoiding partial
1058slabs--, so that freeing ops is identical whether or not this step runs.
057163d7
NC
1059Note that the slab allocator allocates ops downwards in memory, so one would
1060have to actually "allocate" the ops in reverse-execution order to get them
1061contiguous in memory in execution order.
1062
b4af8972
RB
1063See
1064L<http://www.nntp.perl.org/group/perl.perl5.porters/2007/12/msg131975.html>
057163d7
NC
1065
1066Note that running this copy, and then freeing all the old location ops would
1067cause their slabs to be freed, which would eliminate possible memory wastage if
1068the previous suggestion is implemented, and we swap slabs more frequently.
52960e22 1069
12e06b6f
NC
1070=head2 eliminate incorrect line numbers in warnings
1071
1072This code
1073
1074 use warnings;
1075 my $undef;
f703fc96 1076
12e06b6f
NC
1077 if ($undef == 3) {
1078 } elsif ($undef == 0) {
1079 }
1080
18a16cc5 1081used to produce this output:
12e06b6f
NC
1082
1083 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1084 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1085
18a16cc5
NC
1086where the line of the second warning was misreported - it should be line 5.
1087Rafael fixed this - the problem arose because there was no nextstate OP
1088between the execution of the C<if> and the C<elsif>, hence C<PL_curcop> still
1089reports that the currently executing line is line 4. The solution was to inject
1090a nextstate OPs for each C<elsif>, although it turned out that the nextstate
1091OP needed to be a nulled OP, rather than a live nextstate OP, else other line
1092numbers became misreported. (Jenga!)
12e06b6f
NC
1093
1094The problem is more general than C<elsif> (although the C<elsif> case is the
1095most common and the most confusing). Ideally this code
1096
1097 use warnings;
1098 my $undef;
f703fc96 1099
12e06b6f
NC
1100 my $a = $undef + 1;
1101 my $b
1102 = $undef
1103 + 1;
1104
1105would produce this output
1106
1107 Use of uninitialized value $undef in addition (+) at wrong.pl line 4.
1108 Use of uninitialized value $undef in addition (+) at wrong.pl line 7.
1109
1110(rather than lines 4 and 5), but this would seem to require every OP to carry
1111(at least) line number information.
1112
1113What might work is to have an optional line number in memory just before the
1114BASEOP structure, with a flag bit in the op to say whether it's present.
1115Initially during compile every OP would carry its line number. Then add a late
c5fb089a 1116pass to the optimizer (potentially combined with L</repack the optree>) which
12e06b6f
NC
1117looks at the two ops on every edge of the graph of the execution path. If
1118the line number changes, flags the destination OP with this information.
1119Once all paths are traced, replace every op with the flag with a
1120nextstate-light op (that just updates C<PL_curcop>), which in turn then passes
1121control on to the true op. All ops would then be replaced by variants that
1122do not store the line number. (Which, logically, why it would work best in
1123conjunction with L</repack the optree>, as that is already copying/reallocating
1124all the OPs)
1125
18a16cc5
NC
1126(Although I should note that we're not certain that doing this for the general
1127case is worth it)
1128
52960e22
JC
1129=head2 optimize tail-calls
1130
1131Tail-calls present an opportunity for broadly applicable optimization;
1132anywhere that C<< return foo(...) >> is called, the outer return can
1133be replaced by a goto, and foo will return directly to the outer
1134caller, saving (conservatively) 25% of perl's call&return cost, which
1135is relatively higher than in C. The scheme language is known to do
1136this heavily. B::Concise provides good insight into where this
1137optimization is possible, ie anywhere entersub,leavesub op-sequence
1138occurs.
1139
1140 perl -MO=Concise,-exec,a,b,-main -e 'sub a{ 1 }; sub b {a()}; b(2)'
1141
1142Bottom line on this is probably a new pp_tailcall function which
1143combines the code in pp_entersub, pp_leavesub. This should probably
1144be done 1st in XS, and using B::Generate to patch the new OP into the
1145optrees.
1146
e12cb30b 1147=head2 Add C<0odddd>
0c397127
KW
1148
1149It has been proposed that octal constants be specifiable through the syntax
1150C<0oddddd>, parallel to the existing construct to specify hex constants
1151C<0xddddd>
1152
bf7d9bd8
AC
1153=head2 Revisit the regex super-linear cache code
1154
1155Perl executes regexes using the traditional backtracking algorithm, which
1156makes it possible to implement a variety of powerful pattern-matching
1157features (like embedded code blocks), at the cost of taking exponential time
1158to run on some pathological patterns. The exponential-time problem is
1159mitigated by the I<super-linear cache>, which detects when we're processing
1160such a pathological pattern, and does some additional bookkeeping to avoid
1161much of the work. However, that code has bit-rotted a little; some patterns
1162don't make as much use of it as they should. The proposal is to analyse
1163where the current cache code has problems, and extend it to cover those cases.
1164
1165See also
1166L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2013-01/msg00339.html>
1167
0bdfc961
NC
1168=head1 Big projects
1169
1170Tasks that will get your name mentioned in the description of the "Highlights
f6ca94fb 1171of 5.21.2"
0bdfc961
NC
1172
1173=head2 make ithreads more robust
1174
45a81a90 1175Generally make ithreads more robust.
0bdfc961
NC
1176
1177This task is incremental - even a little bit of work on it will help, and
1178will be greatly appreciated.
1179
07577ec1
FC
1180One bit would be to determine how to clone directory handles on systems
1181without a C<fchdir> function (in sv.c:Perl_dirp_dup).
6c047da7 1182
59c7f7d5
RGS
1183Fix Perl_sv_dup, et al so that threads can return objects.
1184
44a7a252
JV
1185=head1 Tasks for microperl
1186
1187
1188[ Each and every one of these may be obsolete, but they were listed
1189 in the old Todo.micro file]
1190
44a7a252
JV
1191=head2 do away with fork/exec/wait?
1192
1193(system, popen should be enough?)
1194
1195=head2 some of the uconfig.sh really needs to be probed (using cc) in buildtime:
1196
1197(uConfigure? :-) native datatype widths and endianness come to mind
1198