This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
doc nits
[perl5.git] / pod / perltodo.pod
CommitLineData
7711098a
GS
1=head1 NAME
2
3perltodo - Perl TO-DO List
4
5=head1 DESCRIPTION
e50bb9a1 6
722d2a37 7This is a list of wishes for Perl. Send updates to
e50bb9a1
GS
8I<perl5-porters@perl.org>. If you want to work on any of these
9projects, be sure to check the perl5-porters archives for past ideas,
10flames, and propaganda. This will save you time and also prevent you
11from implementing something that Larry has already vetoed. One set
12of archives may be found at:
13
14 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/
15
722d2a37 16=head1 To do during 5.6.x
e50bb9a1 17
722d2a37 18=head2 Support for I/O disciplines
e50bb9a1 19
722d2a37
SC
20C<perlio> provides this, but the interface could be a lot more
21straightforward.
e50bb9a1 22
4b3b956a 23=head2 Autoload bytes.pm
e50bb9a1 24
4b3b956a
JH
25When the lexer sees, for instance, C<bytes::length>, it should
26automatically load the C<bytes> pragma.
27
28=head2 Make "\u{XXXX}" et al work
29
30Danger, Will Robinson! Discussing the semantics of C<"\x{F00}">,
31C<"\xF00"> and C<"\U{F00}"> on P5P I<will> lead to a long and boring
32flamewar.
e50bb9a1 33
c6287c21 34=head2 Create a char *sv_pvprintify(sv, STRLEN *lenp, UV flags)
0562c0e3
JH
35
36For displaying PVs with control characters, embedded nulls, and Unicode.
37This would be useful for printing warnings, or data and regex dumping,
38not_a_number(), and so on.
39
1e54db1a 40Requirements: should handle both byte and UTF-8 strings. isPRINT()
f35392ae 41characters printed as-is, character less than 256 as \xHH, Unicode
0661e9a4
JH
42characters as \x{HHH}. Don't assume ASCII-like, either, get somebody
43on EBCDIC to test the output.
f35392ae
JH
44
45Possible options, controlled by the flags:
0661e9a4 46- whitespace (other than ' ' of isPRINT()) printed as-is
f35392ae
JH
47- use isPRINT_LC() instead of isPRINT()
48- print control characters like this: "\cA"
49- print control characters like this: "^A"
0661e9a4
JH
50- non-PRINTables printed as '.' instead of \xHH
51- use \OOO instead of \xHH
52- use the C/Perl-metacharacters like \n, \t
f35392ae
JH
53- have a maximum length for the produced string (read it from *lenp)
54- append a "..." to the produced string if the maximum length is exceeded
0661e9a4 55- really fancy: print unicode characters as \N{...}
f35392ae 56
1626a787
JH
57NOTE: pv_display(), pv_uni_display(), sv_uni_display() are already
58doing something like the above.
c5fc23ff 59
722d2a37 60=head2 Overloadable regex assertions
e50bb9a1 61
722d2a37
SC
62This may or may not be possible with the current regular expression
63engine. The idea is that, for instance, C<\b> needs to be
64algorithmically computed if you're dealing with Thai text. Hence, the
65B<\b> assertion wants to be overloaded by a function.
e50bb9a1 66
776f8809
JH
67=head2 Unicode
68
69=over 4
70
71=item *
e50bb9a1 72
f34dec15
JH
73Allow for long form of the General Category Properties, e.g
74C<\p{IsOpenPunctuation}>, not just the abbreviated form, e.g.
75C<\p{IsPs}>.
76
77=item *
78
1ac13f9a
JH
79Allow for the metaproperties: C<XID Start>, C<XID Continue>,
80C<NF*_NO>, C<NF*_MAYBE> (require the DerivedCoreProperties and
81DerviceNormalizationProperties files).
f34dec15 82
71d929cb
JH
83There are also multiple value properties still unimplemented:
84C<Numeric Type>, C<East Asian Width>.
f34dec15
JH
85
86=item *
87
722d2a37 88 Case Mappings? http://www.unicode.org/unicode/reports/tr21/
e50bb9a1 89
d8b6afd3
JH
90Mostly implemented (all of 1:1, 1:N, N:1), only the "final sigma"
91and locale-specific rules of SpecCase are not implemented.
ac1256e8 92
776f8809 93=item *
e50bb9a1 94
8d3e8850 95UTF-8 identifier names should probably be canonicalized: NFC?
e50bb9a1 96
20eafb1c
JH
97=item *
98
99UTF-8 in package names and sub names? The first is problematic
8d3e8850 100because of the mapping to pathnames, ditto for the second one if
8f8cf39c
JH
101one does autosplitting, for example. Some of this works already
102in 5.8.0, but essentially it is unsupported. Constructs to consider,
103at the very least:
104
105 use utf8;
106 package UnicodePackage;
107 sub new { bless {}, shift };
108 sub UnicodeMethod1 { ... $_[0]->UnicodeMethod2(...) ... }
109 sub UnicodeMethod2 { ... } # in here caller(0) should contain Unicode
110 ...
111 package main;
112 my $x = UnicodePackage->new;
113 print ref $x, "\n"; # should be Unicode
114 $x->UnicodeMethod1(...);
115 my $y = UnicodeMethod3 UnicodePackage ...;
116
117In the above all I<UnicodeXxx> contain (identifier-worthy) characters
118beyond the code point 255, for example 256. Wherever package/class or
119subroutine names can be returned needs to be checked for Unicodeness.
e50bb9a1 120
776f8809
JH
121=back
122
123See L<perlunicode/UNICODE REGULAR EXPRESSION SUPPORT LEVEL> for what's
f34dec15
JH
124there and what's missing. Almost all of Levels 2 and 3 is missing,
125and as of 5.8.0 not even all of Level 1 is there.
8d3e8850 126They have some tricks Perl doesn't yet implement, such as character
20eafb1c
JH
127class subtraction.
128
129 http://www.unicode.org/unicode/reports/tr18/
776f8809 130
56490ca2 131=head2 Work out exit/die semantics for threads
e50bb9a1 132
97b33923
JH
133There are some suggestions to use for example something like this:
134default to "(thread exiting first will) wait for the other threads
135until up to 60 seconds". Other possibilities:
136
137 use threads wait => 0;
138
139Do not wait.
140
141 use threads wait_for => 10;
142
143Wait up to 10 seconds.
144
145 use threads wait_for => -1;
146
147Wait for ever.
e50bb9a1 148
56490ca2 149http://archive.develooper.com/perl5-porters@perl.org/msg79618.html
dd0afe54 150
b2f9d798 151=head2 Better support for nonpreemptive threading systems like GNU pth
dd0afe54 152
b2f9d798
JH
153To better support nonpreemptive threading systems, perhaps some of the
154blocking functions internally in Perl should do a yield() before a
155blocking call. (Now certain threads tests ({basic,list,thread.t})
156simply do a yield() before they sleep() to give nonpreemptive thread
157implementations a chance).
cfde3649 158
b2f9d798
JH
159In some cases, like the GNU pth, which has replacement functions that
160are nonblocking (pth_select instead of select), maybe Perl should be
161using them instead when built for threading.
e50bb9a1 162
722d2a37 163=head2 Typed lexicals for compiler
e50bb9a1 164
722d2a37 165=head2 Compiler workarounds for Win32
e50bb9a1 166
722d2a37 167=head2 AUTOLOADing in the compiler
e50bb9a1 168
722d2a37 169=head2 Fixing comppadlist when compiling
e50bb9a1 170
722d2a37 171=head2 Cleaning up exported namespace
e50bb9a1 172
722d2a37 173=head2 Complete signal handling
e50bb9a1 174
722d2a37
SC
175Add C<PERL_ASYNC_CHECK> to opcodes which loop; replace C<sigsetjmp> with
176C<sigjmp>; check C<wait> for signal safety.
e50bb9a1 177
722d2a37 178=head2 Out-of-source builds
e50bb9a1 179
722d2a37 180This was done for 5.6.0, but needs reworking for 5.7.x
e50bb9a1 181
722d2a37 182=head2 POSIX realtime support
e50bb9a1 183
722d2a37
SC
184POSIX 1003.1 1996 Edition support--realtime stuff: POSIX semaphores,
185message queues, shared memory, realtime clocks, timers, signals (the
186metaconfig units mostly already exist for these)
e50bb9a1 187
722d2a37 188=head2 UNIX98 support
e50bb9a1 189
722d2a37 190Reader-writer locks, realtime/asynchronous IO
e50bb9a1 191
722d2a37 192=head2 IPv6 Support
e50bb9a1 193
fe854a6f 194There are non-core modules, such as C<Socket6>, but these will need
722d2a37
SC
195integrating when IPv6 actually starts to really happen. See RFC 2292
196and RFC 2553.
e50bb9a1 197
722d2a37 198=head2 Long double conversion
e50bb9a1 199
722d2a37 200Floating point formatting is still causing some weird test failures.
e50bb9a1 201
722d2a37 202=head2 Locales
e50bb9a1 203
722d2a37
SC
204Locales and Unicode interact with each other in unpleasant ways.
205One possible solution would be to adopt/support ICU:
e50bb9a1 206
d085b7b6 207 http://oss.software.ibm.com/icu/index.html
e50bb9a1 208
722d2a37 209=head2 Arithmetic on non-Arabic numerals
e50bb9a1 210
722d2a37 211C<[1234567890]> aren't the only numerals any more.
e50bb9a1 212
722d2a37 213=head2 POSIX Unicode character classes
e50bb9a1 214
97b33923 215(C<[=a=]> for equivalence classes, C<[.ch.]> for collation.)
722d2a37 216These are dependent on Unicode normalization and collation.
e50bb9a1 217
722d2a37 218=head2 Factoring out common suffices/prefices in regexps (trie optimization)
c47ff5f1 219
722d2a37
SC
220Currently, the user has to optimize C<foo|far> and C<foo|goo> into
221C<f(?:oo|ar)> and C<[fg]oo> by hand; this could be done automatically.
e50bb9a1 222
722d2a37 223=head2 Security audit shipped utilities
e50bb9a1 224
722d2a37
SC
225All the code we ship with Perl needs to be sensible about temporary file
226handling, locking, input validation, and so on.
e50bb9a1 227
c8d2171d
JH
228=head2 Sort out the uid-setting mess
229
230Currently there are several problems with the setting of uids ($<, $>
231for the real and effective uids). Firstly, what exactly setuid() call
232gets invoked in which platform is simply a big mess that needs to be
233untangled. Secondly, the effects are apparently not standard across
234platforms, (if you first set $< and then $>, or vice versa, being
666f95b9 235uid == euid == zero, or just euid == zero, or as a normal user, what are
c8d2171d
JH
236the results?). The test suite not (usually) being run as root means
237that these things do not get much testing. Thirdly, there's quite
238often a third uid called saved uid, and Perl has no knowledge of that
239feature in any way. (If one has the saved uid of zero, one can get
240back any real and effective uids.) As an example, to change also the
241saved uid, one needs to set the real and effective uids B<twice>-- in
242most systems, that is: in HP-UX that doesn't seem to work.
666f95b9 243
722d2a37 244=head2 Custom opcodes
e50bb9a1 245
722d2a37
SC
246Have a way to introduce user-defined opcodes without the subroutine call
247overhead of an XSUB; the user should be able to create PP code. Simon
248Cozens has some ideas on this.
e50bb9a1 249
722d2a37 250=head2 DLL Versioning
e50bb9a1 251
d1be9408 252Windows needs a way to know what version of an XS or C<libperl> DLL it's
722d2a37 253loading.
e50bb9a1 254
722d2a37 255=head2 Introduce @( and @)
e50bb9a1 256
722d2a37
SC
257C<$(> may return "foo bar baz". Unfortunately, since groups can
258theoretically have spaces in their names, this could be one, two or
259three groups.
e50bb9a1 260
722d2a37 261=head2 Floating point handling
e50bb9a1 262
722d2a37
SC
263C<NaN> and C<inf> support is particularly troublesome.
264(fp_classify(), fp_class(), fp_class_d(), class(), isinf(),
265isfinite(), finite(), isnormal(), unordered(), <ieeefp.h>,
266<fp_class.h> (there are metaconfig units for all these) (I think),
267fp_setmask(), fp_getmask(), fp_setround(), fp_getround()
268(no metaconfig units yet for these). Don't forget finitel(), fp_classl(),
269fp_class_l(), (yes, both do, unfortunately, exist), and unorderedl().)
e50bb9a1 270
210b36aa 271As of Perl 5.6.1, there is a Perl macro, Perl_isnan().
e50bb9a1 272
722d2a37 273=head2 IV/UV preservation
e50bb9a1 274
722d2a37
SC
275Nicholas Clark has done a lot of work on this, but work is continuing.
276C<+>, C<-> and C<*> work, but guards need to be in place for C<%>, C</>,
277C<&>, C<oct>, C<hex> and C<pack>.
e50bb9a1 278
722d2a37 279=head2 Replace pod2html with something using Pod::Parser
83df6a1d 280
fe854a6f 281The CPAN module C<Marek::Pod::Html> may be a more suitable basis for a
97b33923 282C<pod2html> converter; the current one duplicates the functionality
722d2a37
SC
283abstracted in C<Pod::Parser>, which makes updating the POD language
284difficult.
e50bb9a1 285
722d2a37 286=head2 Automate module testing on CPAN
e50bb9a1 287
722d2a37
SC
288When a new Perl is being beta tested, porters have to manually grab
289their favourite CPAN modules and test them - this should be done
290automatically.
e50bb9a1 291
722d2a37 292=head2 sendmsg and recvmsg
83df6a1d 293
722d2a37
SC
294We have all the other BSD socket functions but these. There are
295metaconfig units for these functions which can be added. To avoid these
296being new opcodes, a solution similar to the way C<sockatmark> was added
297would be preferable. (Autoload the C<IO::whatever> module.)
e50bb9a1 298
722d2a37 299=head2 Rewrite perlre documentation
e50bb9a1 300
722d2a37
SC
301The new-style patterns need full documentation, and the whole document
302needs to be a lot clearer.
e50bb9a1 303
722d2a37 304=head2 Convert example code to IO::Handle filehandles
e50bb9a1 305
722d2a37 306=head2 Document Win32 choices
e50bb9a1 307
722d2a37 308=head2 Check new modules
e50bb9a1 309
722d2a37 310=head2 Make roffitall find pods and libs itself
e50bb9a1 311
722d2a37 312Simon Cozens has done some work on this but it needs a rethink.
e50bb9a1 313
722d2a37 314=head1 To do at some point
e50bb9a1 315
722d2a37
SC
316These are ideas that have been regularly tossed around, that most
317people believe should be done maybe during 5.8.x
e50bb9a1 318
722d2a37 319=head2 Remove regular expression recursion
e50bb9a1 320
722d2a37
SC
321Because the regular expression engine is recursive, badly designed
322expressions can lead to lots of recursion filling up the stack. Ilya
323claims that it is easy to convert the engine to being iterative, but
324this has still not yet been done. There may be a regular expression
325engine hit squad meeting at TPC5.
e50bb9a1 326
722d2a37 327=head2 Memory leaks after failed eval
e50bb9a1 328
722d2a37
SC
329Perl will leak memory if you C<eval "hlagh hlagh hlagh hlagh">. This is
330partially because it attempts to build up an op tree for that code and
331doesn't properly free it. The same goes for non-syntactically-correct
332regular expressions. Hugo looked into this, but decided it needed a
333mark-and-sweep GC implementation.
e50bb9a1 334
722d2a37
SC
335Alan notes that: The basic idea was to extend the parser token stack
336(C<YYSTYPE>) to include a type field so we knew what sort of thing each
210b36aa 337element of the stack was. The F<perly.c> code would then have to be
722d2a37
SC
338postprocessed to record the type of each entry on the stack as it was
339created, and the parser patched so that it could unroll the stack
340properly on error.
e50bb9a1 341
722d2a37
SC
342This is possible to do, but would be pretty messy to implement, as it
343would rely on even more sed hackery in F<perly.fixer>.
e50bb9a1 344
722d2a37 345=head2 bitfields in pack
e50bb9a1 346
722d2a37 347=head2 Cross compilation
e50bb9a1 348
722d2a37 349Make Perl buildable with a cross-compiler. This will play havoc with
da75cd15 350Configure, which needs to know how the target system will respond to
722d2a37
SC
351its tests; maybe C<microperl> will be a good starting point here.
352(Indeed, Bart Schuller reports that he compiled up C<microperl> for
353the Agenda PDA and it works fine.) A really big spanner in the works
354is the bootstrapping build process of Perl: if the filesystem the
355target systems sees is not the same what the build host sees, various
356input, output, and (Perl) library files need to be copied back and forth.
e50bb9a1 357
f86a8bc5
JH
358As of 5.8.0 Configure mostly works for cross-compilation
359(used successfully for iPAQ Linux), miniperl gets built,
360but then building DynaLoader (and other extensions) fails
361since MakeMaker knows nothing of cross-compilation.
362(See INSTALL/Cross-compilation for the state of things.)
363
722d2a37 364=head2 Perl preprocessor / macros
e50bb9a1 365
722d2a37
SC
366Source filters help with this, but do not get us all the way. For
367instance, it should be possible to implement the C<??> operator somehow;
368source filters don't (quite) cut it.
e50bb9a1 369
722d2a37 370=head2 Perl lexer in Perl
a45bd81d 371
722d2a37 372Damian Conway is planning to work on this, but it hasn't happened yet.
e50bb9a1 373
722d2a37 374=head2 Using POSIX calls internally
e50bb9a1 375
210b36aa 376When faced with a BSD vs. SysV -style interface to some library or
722d2a37
SC
377system function, perl's roots show in that it typically prefers the BSD
378interface (but falls back to the SysV one). One example is getpgrp().
379Other examples include C<memcpy> vs. C<bcopy>. There are others, mostly in
210b36aa 380F<pp_sys.c>.
e50bb9a1 381
722d2a37
SC
382Mostly, this item is a suggestion for which way to start a journey into
383an C<#ifdef> forest. It is not primarily a suggestion to eliminate any of
384the C<#ifdef> forests.
e50bb9a1 385
722d2a37
SC
386POSIX calls are perhaps more likely to be portable to unexpected
387architectures. They are also perhaps more likely to be actively
388maintained by a current vendor. They are also perhaps more likely to be
389available in thread-safe versions, if appropriate.
e50bb9a1 390
722d2a37 391=head2 -i rename file when changed
e50bb9a1 392
722d2a37
SC
393It's only necessary to rename a file when inplace editing when the file
394has changed. Detecting a change is perhaps the difficult bit.
e50bb9a1 395
722d2a37 396=head2 All ARGV input should act like E<lt>E<gt>
e50bb9a1 397
2d84a16a
DM
398eg C<read(ARGV, ...)> doesn't currently read across multiple files.
399
722d2a37 400=head2 Support for rerunning debugger
e50bb9a1 401
722d2a37 402There should be a way of restarting the debugger on demand.
e50bb9a1 403
c6287c21
JH
404=head2 Test Suite for the Debugger
405
406The debugger is a complex piece of software and fixing something
407here may inadvertently break something else over there. To tame
408this chaotic behaviour, a test suite is necessary.
409
722d2a37 410=head2 my sub foo { }
c47ff5f1 411
722d2a37
SC
412The basic principle is sound, but there are problems with the semantics
413of self-referential and mutually referential lexical subs: how to
414declare the subs?
c47ff5f1 415
722d2a37 416=head2 One-pass global destruction
c47ff5f1 417
722d2a37
SC
418Sweeping away all the allocated memory in one go is a laudable goal, but
419it's difficult and in most cases, it's easier to let the memory get
420freed by exiting.
e50bb9a1 421
722d2a37 422=head2 Rewrite regexp parser
e50bb9a1 423
722d2a37
SC
424There has been talk recently of rewriting the regular expression parser
425to produce an optree instead of a chain of opcodes; it's unclear whether
426or not this would be a win.
e50bb9a1 427
722d2a37 428=head2 Cache recently used regexps
e50bb9a1 429
722d2a37 430This is to speed up
e50bb9a1 431
722d2a37
SC
432 for my $re (@regexps) {
433 $matched++ if /$re/
434 }
e50bb9a1 435
722d2a37
SC
436C<qr//> already gives us a way of saving compiled regexps, but it should
437be done automatically.
e50bb9a1 438
722d2a37 439=head2 Cross-compilation support
04c70446 440
722d2a37
SC
441Bart Schuller reports that using C<microperl> and a cross-compiler, he
442got Perl working on the Agenda PDA. However, one cannot build a full
443Perl because Configure needs to get the results for the target platform,
444for the host.
e50bb9a1 445
722d2a37 446=head2 Bit-shifting bitvectors
e50bb9a1 447
722d2a37 448Given:
e50bb9a1 449
722d2a37 450 vec($v, 1000, 1) = 1;
e50bb9a1 451
722d2a37 452One should be able to do
e50bb9a1 453
722d2a37 454 $v <<= 1;
e50bb9a1 455
722d2a37 456and have the 999'th bit set.
e50bb9a1 457
722d2a37
SC
458Currently if you try with shift bitvectors you shift the NV/UV, instead
459of the bits in the PV. Not very logical.
e50bb9a1 460
722d2a37 461=head2 debugger pragma
e50bb9a1 462
722d2a37
SC
463The debugger is implemented in Perl in F<perl5db.pl>; turning it into a
464pragma should be easy, but making it work lexically might be more
465difficult. Fiddling with C<$^P> would be necessary.
e50bb9a1 466
722d2a37 467=head2 use less pragma
e50bb9a1 468
722d2a37
SC
469Identify areas where speed/memory tradeoffs can be made and have a hint
470to switch between them.
e50bb9a1 471
722d2a37 472=head2 switch structures
e50bb9a1 473
722d2a37
SC
474Although we have C<Switch.pm> in core, Larry points to the dormant
475C<nswitch> and C<cswitch> ops in F<pp.c>; using these opcodes would be
476much faster.
e50bb9a1 477
722d2a37 478=head2 Cache eval tree
e50bb9a1 479
722d2a37 480=head2 rcatmaybe
e50bb9a1 481
722d2a37 482=head2 Shrink opcode tables
e50bb9a1 483
722d2a37 484=head2 Optimize away @_
e50bb9a1 485
722d2a37 486Look at the "reification" code in C<av.c>
e50bb9a1 487
722d2a37 488=head2 Prototypes versus indirect objects
e50bb9a1 489
722d2a37 490Currently, indirect object syntax bypasses prototype checks.
e50bb9a1 491
210b36aa 492=head2 Install HTML
e50bb9a1 493
722d2a37
SC
494HTML versions of the documentation need to be installed by default; a
495call to C<installhtml> from C<installperl> may be all that's necessary.
e50bb9a1 496
722d2a37 497=head2 Prototype method calls
e50bb9a1 498
722d2a37 499=head2 Return context prototype declarations
e50bb9a1 500
722d2a37 501=head2 magic_setisa
e50bb9a1 502
722d2a37 503=head2 Garbage collection
e50bb9a1 504
722d2a37
SC
505There have been persistent mumblings about putting a mark-and-sweep
506garbage detector into Perl; Alan Burlison has some ideas about this.
e50bb9a1 507
722d2a37 508=head2 IO tutorial
e50bb9a1 509
722d2a37 510Mark-Jason Dominus has the beginnings of one of these.
e50bb9a1 511
722d2a37 512=head2 Rewrite perldoc
e50bb9a1 513
722d2a37
SC
514There are a few suggestions for what to do with C<perldoc>: maybe a
515full-text search, an index function, locating pages on a particular
516high-level subject, and so on.
e50bb9a1 517
3958b146 518=head2 Install .3p manpages
e50bb9a1 519
3958b146 520This is a bone of contention; we can create C<.3p> manpages for each
722d2a37
SC
521built-in function, but should we install them by default? Tcl does this,
522and it clutters up C<apropos>.
e50bb9a1 523
722d2a37 524=head2 Unicode tutorial
e50bb9a1 525
722d2a37 526Simon Cozens promises to do this before he gets old.
e50bb9a1 527
722d2a37 528=head2 Update POSIX.pm for 1003.1-2
3958b146 529
722d2a37 530=head2 Retargetable installation
e50bb9a1 531
722d2a37 532Allow C<@INC> to be changed after Perl is built.
e50bb9a1 533
722d2a37 534=head2 POSIX emulation on non-POSIX systems
e50bb9a1 535
722d2a37
SC
536Make C<POSIX.pm> behave as POSIXly as possible everywhere, meaning we
537have to implement POSIX equivalents for some functions if necessary.
e50bb9a1 538
722d2a37 539=head2 Rename Win32 headers
e50bb9a1 540
722d2a37
SC
541=head2 Finish off lvalue functions
542
543They don't work in the debugger, and they don't work for list or hash
544slices.
e50bb9a1 545
722d2a37 546=head2 Update sprintf documentation
e50bb9a1 547
722d2a37 548Hugo van der Sanden plans to look at this.
e50bb9a1 549
722d2a37 550=head2 Use fchown/fchmod internally
e50bb9a1 551
722d2a37
SC
552This has been done in places, but needs a thorough code review.
553Also fchdir is available in some platforms.
e50bb9a1 554
d45541b3 555=head2 Make v-strings overloaded objects
c5fc23ff 556
d45541b3
JH
557Instead of having to guess whether a string is a v-string and thus
558needs to be displayed with %vd, make v-strings (readonly) objects
559(class "vstring"?) with a stringify overload.
c5fc23ff 560
49293501
MS
561=head2 Allow restricted hash assignment
562
563Currently you're not allowed to assign to a restricted hash at all,
564even with the same keys.
565
566 %restricted = (foo => 42); # error
567
568This should be allowed if the new keyset is a subset of the old
569keyset. May require more extra code than we'd like in pp_aassign.
570
5387ccf1
JH
571=head2 Should overload be inheritable?
572
573Should overload be 'contagious' through @ISA so that derived classes
574would inherit their base classes' overload definitions? What to do
575in case of overload conflicts?
576
cbda53d5
JH
577=head2 Taint rethink
578
579Should taint be stopped from affecting control flow, if ($tainted)?
580Should tainted symbolic method calls and subref calls be stopped?
581(Look at Ruby's $SAFE levels for inspiration?)
582
7619c85e
RG
583=head2 Perform correctly when XSUBs call subroutines that exit via goto(LABEL) and friends
584
585If an XSUB calls a subroutine that exits using goto(LABEL),
586last(LABEL) or next(LABEL), then the interpreter will very probably crash
587with a segfault because the execution resumes in the XSUB instead of
588never returning there.
589
722d2a37 590=head1 Vague ideas
e50bb9a1 591
722d2a37 592Ideas which have been discussed, and which may or may not happen.
e50bb9a1 593
722d2a37 594=head2 ref() in list context
e50bb9a1 595
722d2a37
SC
596It's unclear what this should do or how to do it without breaking old
597code.
e50bb9a1 598
f86a8bc5 599=head2 Make tr/// return histogram of characters in list context
e50bb9a1 600
722d2a37 601There is a patch for this, but it may require Unicodification.
e50bb9a1 602
722d2a37 603=head2 Compile to real threaded code
3958b146 604
722d2a37 605=head2 Structured types
3958b146 606
722d2a37 607=head2 Modifiable $1 et al.
e50bb9a1 608
722d2a37
SC
609 ($x = "elephant") =~ /e(ph)/;
610 $1 = "g"; # $x = "elegant"
e50bb9a1 611
722d2a37
SC
612What happens if there are multiple (nested?) brackets? What if the
613string changes between the match and the assignment?
e50bb9a1 614
722d2a37 615=head2 Procedural interfaces for IO::*, etc.
e50bb9a1 616
722d2a37
SC
617Some core modules have been accused of being overly-OO. Adding
618procedural interfaces could demystify them.
e50bb9a1 619
722d2a37 620=head2 RPC modules
e50bb9a1 621
722d2a37 622=head2 Attach/detach debugger from running program
e50bb9a1 623
722d2a37
SC
624With C<gdb>, you can attach the debugger to a running program if you
625pass the process ID. It would be good to do this with the Perl debugger
626on a running Perl program, although I'm not sure how it would be done.
e50bb9a1 627
722d2a37 628=head2 GUI::Native
e50bb9a1 629
722d2a37
SC
630A non-core module that would use "native" GUI to create graphical
631applications.
e50bb9a1 632
722d2a37 633=head2 foreach(reverse ...)
e50bb9a1 634
722d2a37 635Currently
e50bb9a1 636
722d2a37 637 foreach (reverse @_) { ... }
e50bb9a1 638
722d2a37
SC
639puts C<@_> on the stack, reverses it putting the reversed version on the
640stack, then iterates forwards. Instead, it could be special-cased to put
641C<@_> on the stack then iterate backwards.
e50bb9a1 642
722d2a37 643=head2 Constant function cache
e50bb9a1 644
722d2a37 645=head2 Approximate regular expression matching
e50bb9a1 646
722d2a37 647=head1 Ongoing
e50bb9a1 648
722d2a37 649These items B<always> need doing:
e50bb9a1 650
722d2a37 651=head2 Update guts documentation
e50bb9a1 652
722d2a37
SC
653Simon Cozens tries to do this when possible, and contributions to the
654C<perlapi> documentation is welcome.
e50bb9a1 655
722d2a37 656=head2 Add more tests
e50bb9a1 657
722d2a37
SC
658Michael Schwern will donate $500 to Yet Another Society when all core
659modules have tests.
e50bb9a1 660
722d2a37 661=head2 Update auxiliary tools
e50bb9a1 662
722d2a37 663The code we ship with Perl should look like good Perl 5.
e50bb9a1 664
1e278fd9
JH
665=head2 Create debugging macros
666
667Debugging macros (like printsv, dump) can make debugging perl inside a
668C debugger much easier. A good set for gdb comes with mod_perl.
669Something similar should be distributed with perl.
670
671The proper way to do this is to use and extend Devel::DebugInit.
672Devel::DebugInit also needs to be extended to support threads.
673
674See p5p archives for late May/early June 2001 for a recent discussion
675on this topic.
676
677=head2 truncate to the people
678
679One can emulate ftruncate() using F_FREESP and F_CHSIZ fcntls
680(see the UNIX FAQ for details). This needs to go somewhere near
681pp_sys.c:pp_truncate().
682
683One can emulate truncate() easily if one has ftruncate().
684This emulation should also go near pp_sys.pp_truncate().
685
686=head2 Unicode in Filenames
687
613693f3
JH
688chdir, chmod, chown, chroot, exec, glob, link, lstat, mkdir, open,
689opendir, qx, readdir, readlink, rename, rmdir, stat, symlink, sysopen,
1e8e8236 690system, truncate, unlink, utime, -X. All these could potentially accept
613693f3
JH
691Unicode filenames either as input or output (and in the case of system
692and qx Unicode in general, as input or output to/from the shell).
693Whether a filesystem - an operating system pair understands Unicode in
694filenames varies.
1e278fd9
JH
695
696Known combinations that have some level of understanding include
697Microsoft NTFS, Apple HFS+ (In Mac OS 9 and X) and Apple UFS (in Mac
698OS X), NFS v4 is rumored to be Unicode, and of course Plan 9. How to
699create Unicode filenames, what forms of Unicode are accepted and used
700(UCS-2, UTF-16, UTF-8), what (if any) is the normalization form used,
701and so on, varies. Finding the right level of interfacing to Perl
702requires some thought. Remember that an OS does not implicate a
703filesystem.
704
1aad1664
JH
705(The Windows -C command flag "wide API support" has been at least
706temporarily retired in 5.8.1, and the -C has been repurposed, see
707L<perlrun>.)
708
709=head1 Unicode in %ENV
710
711Currently the %ENV entries are always byte strings.
eb450546 712
722d2a37 713=head1 Recently done things
e50bb9a1 714
722d2a37
SC
715These are things which have been on the todo lists in previous releases
716but have recently been completed.
e50bb9a1 717
b0b7f283 718=head2 Alternative RE syntax module
719
720The C<Regexp::English> module, available from the CPAN, provides this:
721
722 my $re = Regexp::English
723 -> start_of_line
724 -> literal('Flippers')
725 -> literal(':')
726 -> optional
727 -> whitespace_char
728 -> end
729 -> remember
730 -> multiple
731 -> digit;
732
733 /$re/;
734
722d2a37 735=head2 Safe signal handling
e50bb9a1 736
722d2a37
SC
737A new signal model went into 5.7.1 without much fanfare. Operations and
738C<malloc>s are no longer interrupted by signals, which are handled
739between opcodes. This means that C<PERL_ASYNC_CHECK> now actually does
740something. However, there are still a few things that need to be done.
e50bb9a1 741
722d2a37 742=head2 Tie Modules
e50bb9a1 743
722d2a37
SC
744Modules which implement arrays in terms of strings, substrings or files
745can be found on the CPAN.
e50bb9a1 746
722d2a37 747=head2 gettimeofday
e50bb9a1 748
210b36aa 749C<Time::HiRes> has been integrated into the core.
e50bb9a1 750
722d2a37 751=head2 setitimer and getimiter
e50bb9a1 752
210b36aa 753Adding C<Time::HiRes> got us this too.
e50bb9a1 754
722d2a37
SC
755=head2 Testing __DIE__ hook
756
757Tests have been added.
758
759=head2 CPP equivalent in Perl
e50bb9a1 760
722d2a37
SC
761A C Yardley will probably have done this by the time you can read this.
762This allows for a generalization of the C constant detection used in
763building C<Errno.pm>.
e50bb9a1 764
722d2a37 765=head2 Explicit switch statements
e50bb9a1 766
722d2a37
SC
767C<Switch.pm> has been integrated into the core to give you all manner of
768C<switch...case> semantics.
e50bb9a1 769
722d2a37 770=head2 autocroak
e50bb9a1 771
722d2a37 772This is C<Fatal.pm>.
e50bb9a1 773
722d2a37 774=head2 UTF/EBCDIC
e50bb9a1 775
722d2a37 776Nick Ing-Simmons has made UTF-EBCDIC (UTR13) work with Perl.
e50bb9a1 777
722d2a37 778 EBCDIC? http://www.unicode.org/unicode/reports/tr16/
e50bb9a1 779
722d2a37 780=head2 UTF Regexes
e50bb9a1 781
722d2a37
SC
782Although there are probably some small bugs to be rooted out, Jarkko
783Hietaniemi has made regular expressions polymorphic between bytes and
784characters.
e50bb9a1 785
722d2a37 786=head2 perlcc to produce executable
e50bb9a1 787
722d2a37
SC
788C<perlcc> was recently rewritten, and can now produce standalone
789executables.
e50bb9a1 790
722d2a37 791=head2 END blocks saved in compiled output
e50bb9a1 792
722d2a37 793=head2 Secure temporary file module
e50bb9a1 794
722d2a37 795Tim Jenness' C<File::Temp> is now in core.
e50bb9a1 796
722d2a37 797=head2 Integrate Time::HiRes
e50bb9a1 798
722d2a37 799This module is now part of core.
e50bb9a1 800
722d2a37 801=head2 Turn Cwd into XS
e50bb9a1 802
722d2a37 803Benjamin Sugars has done this.
e50bb9a1 804
722d2a37 805=head2 Mmap for input
e50bb9a1 806
722d2a37 807Nick Ing-Simmons' C<perlio> supports an C<mmap> IO method.
e50bb9a1 808
1e54db1a 809=head2 Byte to/from UTF-8 and UTF-8 to/from local conversion
e50bb9a1 810
722d2a37 811C<Encode> provides this.
e50bb9a1 812
722d2a37 813=head2 Add sockatmark support
e50bb9a1 814
722d2a37 815Added in 5.7.1
e50bb9a1 816
722d2a37
SC
817=head2 Mailing list archives
818
f224927c 819http://lists.perl.org/ , http://archive.develooper.com/
722d2a37
SC
820
821=head2 Bug tracking
822
ee8e686d
RF
823Since 5.8.0 perl uses the RT bug tracking system from Jesse Vincent,
824implemented by Robert Spier at http://bugs.perl.org/
e50bb9a1 825
722d2a37 826=head2 Integrate MacPerl
e50bb9a1 827
722d2a37
SC
828Chris Nandor and Matthias Neeracher have integrated the MacPerl changes
829into 5.6.0.
e50bb9a1 830
722d2a37 831=head2 Web "nerve center" for Perl
e50bb9a1 832
722d2a37 833http://use.perl.org/ is what you're looking for.
e50bb9a1 834
722d2a37 835=head2 Regular expression tutorial
e50bb9a1 836
722d2a37 837C<perlretut>, provided by Mark Kvale.
e50bb9a1 838
722d2a37 839=head2 Debugging Tutorial
e50bb9a1 840
722d2a37 841C<perldebtut>, written by Richard Foley.
e50bb9a1 842
722d2a37 843=head2 Integrate new modules
e50bb9a1 844
722d2a37 845Jarkko has been integrating madly into 5.7.x
e50bb9a1 846
722d2a37 847=head2 Integrate profiler
e50bb9a1 848
722d2a37 849C<Devel::DProf> is now a core module.
e50bb9a1 850
722d2a37 851=head2 Y2K error detection
e50bb9a1 852
722d2a37
SC
853There's a configure option to detect unsafe concatenation with "19", and
854a CPAN module. (C<D'oh::Year>)
e50bb9a1 855
722d2a37 856=head2 Regular expression debugger
e50bb9a1 857
722d2a37
SC
858While not part of core, Mark-Jason Dominus has written C<Rx> and has
859also come up with a generalised strategy for regular expression
860debugging.
e50bb9a1 861
722d2a37 862=head2 POD checker
e50bb9a1 863
722d2a37 864That's, uh, F<podchecker>
e50bb9a1 865
722d2a37 866=head2 "Dynamic" lexicals
e50bb9a1 867
722d2a37 868=head2 Cache precompiled modules
e50bb9a1 869
722d2a37 870=head1 Deprecated Wishes
e50bb9a1 871
722d2a37
SC
872These are items which used to be in the todo file, but have been
873deprecated for some reason.
e50bb9a1 874
722d2a37 875=head2 Loop control on do{}
e50bb9a1 876
722d2a37 877This would break old code; use C<do{{ }}> instead.
e50bb9a1 878
722d2a37 879=head2 Lexically scoped typeglobs
e50bb9a1 880
722d2a37 881Not needed now we have lexical IO handles.
e50bb9a1 882
722d2a37 883=head2 format BOTTOM
3958b146 884
722d2a37 885=head2 report HANDLE
e50bb9a1 886
722d2a37 887Damian Conway's text formatting modules seem to be the Way To Go.
e50bb9a1 888
722d2a37 889=head2 Generalised want()/caller())
3958b146 890
638ae6a9
MJD
891Robin Houston's C<Want> module does this.
892
722d2a37 893=head2 Named prototypes
e50bb9a1 894
638ae6a9 895This seems to be delayed until Perl 6.
e50bb9a1 896
722d2a37 897=head2 Built-in globbing
e50bb9a1 898
722d2a37 899The C<File::Glob> module has been used to replace the C<glob> function.
e50bb9a1 900
722d2a37 901=head2 Regression tests for suidperl
e50bb9a1 902
722d2a37 903C<suidperl> is deprecated in favour of common sense.
e50bb9a1 904
722d2a37 905=head2 Cached hash values
e50bb9a1 906
722d2a37 907We have shared hash keys, which perform the same job.
e50bb9a1 908
722d2a37 909=head2 Add compression modules
e50bb9a1 910
722d2a37
SC
911The compression modules are a little heavy; meanwhile, Nick Clark is
912working on experimental pragmata to do transparent decompression on
913input.
e50bb9a1 914
722d2a37 915=head2 Reorganise documentation into tutorials/references
e50bb9a1 916
722d2a37 917Could not get consensus on P5P about this.
e50bb9a1 918
722d2a37
SC
919=head2 Remove distinction between functions and operators
920
921Caution: highly flammable.
922
923=head2 Make XS easier to use
e50bb9a1 924
722d2a37 925Use C<Inline> instead, or SWIG.
e50bb9a1 926
722d2a37 927=head2 Make embedding easier to use
e50bb9a1 928
722d2a37 929Use C<Inline::CPR>.
e50bb9a1 930
722d2a37 931=head2 man for perl
04c70446 932
1577cd80 933See the Perl Power Tools. ( http://language.perl.com/ppt/ )
04c70446 934
722d2a37 935=head2 my $Package::variable
04c70446 936
722d2a37 937Use C<our> instead.
04c70446 938
722d2a37 939=head2 "or" tests defined, not truth
04c70446 940
722d2a37 941Suggesting this on P5P B<will> cause a boring and interminable flamewar.
04c70446 942
722d2a37 943=head2 "class"-based lexicals
04c70446 944
cbb3fa72 945Use flyweight objects, secure hashes or, dare I say it, pseudo-hashes instead.
f86a8bc5 946(Or whatever will replace pseudohashes in 5.10.)
04c70446 947
722d2a37 948=head2 byteperl
04c70446 949
722d2a37 950C<ByteLoader> covers this.
04c70446 951
722d2a37 952=head2 Lazy evaluation / tail recursion removal
04c70446 953
f86a8bc5
JH
954C<List::Util> gives first() (a short-circuiting grep); tail recursion
955removal is done manually, with C<goto &whoami;>. (However, MJD has
956found that C<goto &whoami> introduces a performance penalty, so maybe
957there should be a way to do this after all: C<sub foo {START: ... goto
958START;> is better.)
0562c0e3
JH
959
960=head2 Make "use utf8" the default
961
f86a8bc5
JH
962Because of backward compatibility this is difficult: scripts could not
963contain B<any legacy eight-bit data> (like Latin-1) anymore, even in
964string literals or pod. Also would introduce a measurable slowdown of
965at least few percentages since all regular expression operations would
966be done in full UTF-8. But if you want to try this, add
967-DUSE_UTF8_SCRIPTS to your compilation flags.
968
3298bd4d
JH
969=head2 Unicode collation and normalization
970
971The Unicode::Collate and Unicode::Normalize modules
972by SADAHIRO Tomoyuki have been included since 5.8.0.
973
974 Collation? http://www.unicode.org/unicode/reports/tr10/
975 Normalization? http://www.unicode.org/unicode/reports/tr15/
0562c0e3 976
1626a787
JH
977=head2 pack/unpack tutorial
978
979Wolfgang Laun finished what Simon Cozens started.
980
3298bd4d 981=cut