Commit | Line | Data |
7711098a |
1 | =head1 NAME |
2 | |
3 | perltodo - Perl TO-DO List |
4 | |
5 | =head1 DESCRIPTION |
e50bb9a1 |
6 | |
722d2a37 |
7 | This is a list of wishes for Perl. Send updates to |
e50bb9a1 |
8 | I<perl5-porters@perl.org>. If you want to work on any of these |
9 | projects, be sure to check the perl5-porters archives for past ideas, |
10 | flames, and propaganda. This will save you time and also prevent you |
11 | from implementing something that Larry has already vetoed. One set |
12 | of 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 |
20 | C<perlio> provides this, but the interface could be a lot more |
21 | straightforward. |
e50bb9a1 |
22 | |
4b3b956a |
23 | =head2 Autoload bytes.pm |
e50bb9a1 |
24 | |
4b3b956a |
25 | When the lexer sees, for instance, C<bytes::length>, it should |
26 | automatically load the C<bytes> pragma. |
27 | |
28 | =head2 Make "\u{XXXX}" et al work |
29 | |
30 | Danger, Will Robinson! Discussing the semantics of C<"\x{F00}">, |
31 | C<"\xF00"> and C<"\U{F00}"> on P5P I<will> lead to a long and boring |
32 | flamewar. |
e50bb9a1 |
33 | |
c6287c21 |
34 | =head2 Create a char *sv_pvprintify(sv, STRLEN *lenp, UV flags) |
0562c0e3 |
35 | |
36 | For displaying PVs with control characters, embedded nulls, and Unicode. |
37 | This would be useful for printing warnings, or data and regex dumping, |
38 | not_a_number(), and so on. |
39 | |
f35392ae |
40 | Requirements: should handle both byte and UTF8 strings. isPRINT() |
41 | characters printed as-is, character less than 256 as \xHH, Unicode |
0661e9a4 |
42 | characters as \x{HHH}. Don't assume ASCII-like, either, get somebody |
43 | on EBCDIC to test the output. |
f35392ae |
44 | |
45 | Possible options, controlled by the flags: |
0661e9a4 |
46 | - whitespace (other than ' ' of isPRINT()) printed as-is |
f35392ae |
47 | - use isPRINT_LC() instead of isPRINT() |
48 | - print control characters like this: "\cA" |
49 | - print control characters like this: "^A" |
0661e9a4 |
50 | - non-PRINTables printed as '.' instead of \xHH |
51 | - use \OOO instead of \xHH |
52 | - use the C/Perl-metacharacters like \n, \t |
f35392ae |
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 |
57 | NOTE: pv_display(), pv_uni_display(), sv_uni_display() are already |
58 | doing something like the above. |
c5fc23ff |
59 | |
722d2a37 |
60 | =head2 Overloadable regex assertions |
e50bb9a1 |
61 | |
722d2a37 |
62 | This may or may not be possible with the current regular expression |
63 | engine. The idea is that, for instance, C<\b> needs to be |
64 | algorithmically computed if you're dealing with Thai text. Hence, the |
65 | B<\b> assertion wants to be overloaded by a function. |
e50bb9a1 |
66 | |
776f8809 |
67 | =head2 Unicode |
68 | |
69 | =over 4 |
70 | |
71 | =item * |
e50bb9a1 |
72 | |
f34dec15 |
73 | Allow for long form of the General Category Properties, e.g |
74 | C<\p{IsOpenPunctuation}>, not just the abbreviated form, e.g. |
75 | C<\p{IsPs}>. |
76 | |
77 | =item * |
78 | |
1ac13f9a |
79 | Allow for the metaproperties: C<XID Start>, C<XID Continue>, |
80 | C<NF*_NO>, C<NF*_MAYBE> (require the DerivedCoreProperties and |
81 | DerviceNormalizationProperties files). |
f34dec15 |
82 | |
71d929cb |
83 | There are also multiple value properties still unimplemented: |
84 | C<Numeric Type>, C<East Asian Width>. |
f34dec15 |
85 | |
86 | =item * |
87 | |
722d2a37 |
88 | Case Mappings? http://www.unicode.org/unicode/reports/tr21/ |
e50bb9a1 |
89 | |
6f16a292 |
90 | lc(), uc(), lcfirst(), and ucfirst() work only for some of the |
91 | simplest cases, where the mapping goes from a single Unicode character |
92 | to another single Unicode character. See lib/unicore/SpecCase.txt |
93 | (and CaseFold.txt). |
ac1256e8 |
94 | |
776f8809 |
95 | =item * |
e50bb9a1 |
96 | |
8d3e8850 |
97 | UTF-8 identifier names should probably be canonicalized: NFC? |
e50bb9a1 |
98 | |
20eafb1c |
99 | =item * |
100 | |
101 | UTF-8 in package names and sub names? The first is problematic |
8d3e8850 |
102 | because of the mapping to pathnames, ditto for the second one if |
20eafb1c |
103 | one does autosplitting, for example. |
e50bb9a1 |
104 | |
776f8809 |
105 | =back |
106 | |
107 | See L<perlunicode/UNICODE REGULAR EXPRESSION SUPPORT LEVEL> for what's |
f34dec15 |
108 | there and what's missing. Almost all of Levels 2 and 3 is missing, |
109 | and as of 5.8.0 not even all of Level 1 is there. |
8d3e8850 |
110 | They have some tricks Perl doesn't yet implement, such as character |
20eafb1c |
111 | class subtraction. |
112 | |
113 | http://www.unicode.org/unicode/reports/tr18/ |
776f8809 |
114 | |
56490ca2 |
115 | =head2 Work out exit/die semantics for threads |
e50bb9a1 |
116 | |
97b33923 |
117 | There are some suggestions to use for example something like this: |
118 | default to "(thread exiting first will) wait for the other threads |
119 | until up to 60 seconds". Other possibilities: |
120 | |
121 | use threads wait => 0; |
122 | |
123 | Do not wait. |
124 | |
125 | use threads wait_for => 10; |
126 | |
127 | Wait up to 10 seconds. |
128 | |
129 | use threads wait_for => -1; |
130 | |
131 | Wait for ever. |
e50bb9a1 |
132 | |
56490ca2 |
133 | http://archive.develooper.com/perl5-porters@perl.org/msg79618.html |
dd0afe54 |
134 | |
56490ca2 |
135 | =head2 Better support for nonpreemptive threading systems like PTH |
dd0afe54 |
136 | |
56490ca2 |
137 | Perhaps do yield() before blocking calls, exchange calls to non blocking |
138 | calls specific for those libraries (for example GNUpth has a pth_sleep |
139 | which is a non pth aware version of sleep). |
e50bb9a1 |
140 | |
722d2a37 |
141 | =head2 Typed lexicals for compiler |
e50bb9a1 |
142 | |
722d2a37 |
143 | =head2 Compiler workarounds for Win32 |
e50bb9a1 |
144 | |
722d2a37 |
145 | =head2 AUTOLOADing in the compiler |
e50bb9a1 |
146 | |
722d2a37 |
147 | =head2 Fixing comppadlist when compiling |
e50bb9a1 |
148 | |
722d2a37 |
149 | =head2 Cleaning up exported namespace |
e50bb9a1 |
150 | |
722d2a37 |
151 | =head2 Complete signal handling |
e50bb9a1 |
152 | |
722d2a37 |
153 | Add C<PERL_ASYNC_CHECK> to opcodes which loop; replace C<sigsetjmp> with |
154 | C<sigjmp>; check C<wait> for signal safety. |
e50bb9a1 |
155 | |
722d2a37 |
156 | =head2 Out-of-source builds |
e50bb9a1 |
157 | |
722d2a37 |
158 | This was done for 5.6.0, but needs reworking for 5.7.x |
e50bb9a1 |
159 | |
722d2a37 |
160 | =head2 POSIX realtime support |
e50bb9a1 |
161 | |
722d2a37 |
162 | POSIX 1003.1 1996 Edition support--realtime stuff: POSIX semaphores, |
163 | message queues, shared memory, realtime clocks, timers, signals (the |
164 | metaconfig units mostly already exist for these) |
e50bb9a1 |
165 | |
722d2a37 |
166 | =head2 UNIX98 support |
e50bb9a1 |
167 | |
722d2a37 |
168 | Reader-writer locks, realtime/asynchronous IO |
e50bb9a1 |
169 | |
722d2a37 |
170 | =head2 IPv6 Support |
e50bb9a1 |
171 | |
fe854a6f |
172 | There are non-core modules, such as C<Socket6>, but these will need |
722d2a37 |
173 | integrating when IPv6 actually starts to really happen. See RFC 2292 |
174 | and RFC 2553. |
e50bb9a1 |
175 | |
722d2a37 |
176 | =head2 Long double conversion |
e50bb9a1 |
177 | |
722d2a37 |
178 | Floating point formatting is still causing some weird test failures. |
e50bb9a1 |
179 | |
722d2a37 |
180 | =head2 Locales |
e50bb9a1 |
181 | |
722d2a37 |
182 | Locales and Unicode interact with each other in unpleasant ways. |
183 | One possible solution would be to adopt/support ICU: |
e50bb9a1 |
184 | |
722d2a37 |
185 | http://oss.software.ibm.com/developerworks/opensource/icu/project/ |
e50bb9a1 |
186 | |
722d2a37 |
187 | =head2 Arithmetic on non-Arabic numerals |
e50bb9a1 |
188 | |
722d2a37 |
189 | C<[1234567890]> aren't the only numerals any more. |
e50bb9a1 |
190 | |
722d2a37 |
191 | =head2 POSIX Unicode character classes |
e50bb9a1 |
192 | |
97b33923 |
193 | (C<[=a=]> for equivalence classes, C<[.ch.]> for collation.) |
722d2a37 |
194 | These are dependent on Unicode normalization and collation. |
e50bb9a1 |
195 | |
722d2a37 |
196 | =head2 Factoring out common suffices/prefices in regexps (trie optimization) |
c47ff5f1 |
197 | |
722d2a37 |
198 | Currently, the user has to optimize C<foo|far> and C<foo|goo> into |
199 | C<f(?:oo|ar)> and C<[fg]oo> by hand; this could be done automatically. |
e50bb9a1 |
200 | |
722d2a37 |
201 | =head2 Security audit shipped utilities |
e50bb9a1 |
202 | |
722d2a37 |
203 | All the code we ship with Perl needs to be sensible about temporary file |
204 | handling, locking, input validation, and so on. |
e50bb9a1 |
205 | |
c8d2171d |
206 | =head2 Sort out the uid-setting mess |
207 | |
208 | Currently there are several problems with the setting of uids ($<, $> |
209 | for the real and effective uids). Firstly, what exactly setuid() call |
210 | gets invoked in which platform is simply a big mess that needs to be |
211 | untangled. Secondly, the effects are apparently not standard across |
212 | platforms, (if you first set $< and then $>, or vice versa, being |
666f95b9 |
213 | uid == euid == zero, or just euid == zero, or as a normal user, what are |
c8d2171d |
214 | the results?). The test suite not (usually) being run as root means |
215 | that these things do not get much testing. Thirdly, there's quite |
216 | often a third uid called saved uid, and Perl has no knowledge of that |
217 | feature in any way. (If one has the saved uid of zero, one can get |
218 | back any real and effective uids.) As an example, to change also the |
219 | saved uid, one needs to set the real and effective uids B<twice>-- in |
220 | most systems, that is: in HP-UX that doesn't seem to work. |
666f95b9 |
221 | |
722d2a37 |
222 | =head2 Custom opcodes |
e50bb9a1 |
223 | |
722d2a37 |
224 | Have a way to introduce user-defined opcodes without the subroutine call |
225 | overhead of an XSUB; the user should be able to create PP code. Simon |
226 | Cozens has some ideas on this. |
e50bb9a1 |
227 | |
722d2a37 |
228 | =head2 DLL Versioning |
e50bb9a1 |
229 | |
d1be9408 |
230 | Windows needs a way to know what version of an XS or C<libperl> DLL it's |
722d2a37 |
231 | loading. |
e50bb9a1 |
232 | |
722d2a37 |
233 | =head2 Introduce @( and @) |
e50bb9a1 |
234 | |
722d2a37 |
235 | C<$(> may return "foo bar baz". Unfortunately, since groups can |
236 | theoretically have spaces in their names, this could be one, two or |
237 | three groups. |
e50bb9a1 |
238 | |
722d2a37 |
239 | =head2 Floating point handling |
e50bb9a1 |
240 | |
722d2a37 |
241 | C<NaN> and C<inf> support is particularly troublesome. |
242 | (fp_classify(), fp_class(), fp_class_d(), class(), isinf(), |
243 | isfinite(), finite(), isnormal(), unordered(), <ieeefp.h>, |
244 | <fp_class.h> (there are metaconfig units for all these) (I think), |
245 | fp_setmask(), fp_getmask(), fp_setround(), fp_getround() |
246 | (no metaconfig units yet for these). Don't forget finitel(), fp_classl(), |
247 | fp_class_l(), (yes, both do, unfortunately, exist), and unorderedl().) |
e50bb9a1 |
248 | |
210b36aa |
249 | As of Perl 5.6.1, there is a Perl macro, Perl_isnan(). |
e50bb9a1 |
250 | |
722d2a37 |
251 | =head2 IV/UV preservation |
e50bb9a1 |
252 | |
722d2a37 |
253 | Nicholas Clark has done a lot of work on this, but work is continuing. |
254 | C<+>, C<-> and C<*> work, but guards need to be in place for C<%>, C</>, |
255 | C<&>, C<oct>, C<hex> and C<pack>. |
e50bb9a1 |
256 | |
722d2a37 |
257 | =head2 Replace pod2html with something using Pod::Parser |
83df6a1d |
258 | |
fe854a6f |
259 | The CPAN module C<Marek::Pod::Html> may be a more suitable basis for a |
97b33923 |
260 | C<pod2html> converter; the current one duplicates the functionality |
722d2a37 |
261 | abstracted in C<Pod::Parser>, which makes updating the POD language |
262 | difficult. |
e50bb9a1 |
263 | |
722d2a37 |
264 | =head2 Automate module testing on CPAN |
e50bb9a1 |
265 | |
722d2a37 |
266 | When a new Perl is being beta tested, porters have to manually grab |
267 | their favourite CPAN modules and test them - this should be done |
268 | automatically. |
e50bb9a1 |
269 | |
722d2a37 |
270 | =head2 sendmsg and recvmsg |
83df6a1d |
271 | |
722d2a37 |
272 | We have all the other BSD socket functions but these. There are |
273 | metaconfig units for these functions which can be added. To avoid these |
274 | being new opcodes, a solution similar to the way C<sockatmark> was added |
275 | would be preferable. (Autoload the C<IO::whatever> module.) |
e50bb9a1 |
276 | |
722d2a37 |
277 | =head2 Rewrite perlre documentation |
e50bb9a1 |
278 | |
722d2a37 |
279 | The new-style patterns need full documentation, and the whole document |
280 | needs to be a lot clearer. |
e50bb9a1 |
281 | |
722d2a37 |
282 | =head2 Convert example code to IO::Handle filehandles |
e50bb9a1 |
283 | |
722d2a37 |
284 | =head2 Document Win32 choices |
e50bb9a1 |
285 | |
722d2a37 |
286 | =head2 Check new modules |
e50bb9a1 |
287 | |
722d2a37 |
288 | =head2 Make roffitall find pods and libs itself |
e50bb9a1 |
289 | |
722d2a37 |
290 | Simon Cozens has done some work on this but it needs a rethink. |
e50bb9a1 |
291 | |
722d2a37 |
292 | =head1 To do at some point |
e50bb9a1 |
293 | |
722d2a37 |
294 | These are ideas that have been regularly tossed around, that most |
295 | people believe should be done maybe during 5.8.x |
e50bb9a1 |
296 | |
722d2a37 |
297 | =head2 Remove regular expression recursion |
e50bb9a1 |
298 | |
722d2a37 |
299 | Because the regular expression engine is recursive, badly designed |
300 | expressions can lead to lots of recursion filling up the stack. Ilya |
301 | claims that it is easy to convert the engine to being iterative, but |
302 | this has still not yet been done. There may be a regular expression |
303 | engine hit squad meeting at TPC5. |
e50bb9a1 |
304 | |
722d2a37 |
305 | =head2 Memory leaks after failed eval |
e50bb9a1 |
306 | |
722d2a37 |
307 | Perl will leak memory if you C<eval "hlagh hlagh hlagh hlagh">. This is |
308 | partially because it attempts to build up an op tree for that code and |
309 | doesn't properly free it. The same goes for non-syntactically-correct |
310 | regular expressions. Hugo looked into this, but decided it needed a |
311 | mark-and-sweep GC implementation. |
e50bb9a1 |
312 | |
722d2a37 |
313 | Alan notes that: The basic idea was to extend the parser token stack |
314 | (C<YYSTYPE>) to include a type field so we knew what sort of thing each |
210b36aa |
315 | element of the stack was. The F<perly.c> code would then have to be |
722d2a37 |
316 | postprocessed to record the type of each entry on the stack as it was |
317 | created, and the parser patched so that it could unroll the stack |
318 | properly on error. |
e50bb9a1 |
319 | |
722d2a37 |
320 | This is possible to do, but would be pretty messy to implement, as it |
321 | would rely on even more sed hackery in F<perly.fixer>. |
e50bb9a1 |
322 | |
722d2a37 |
323 | =head2 bitfields in pack |
e50bb9a1 |
324 | |
722d2a37 |
325 | =head2 Cross compilation |
e50bb9a1 |
326 | |
722d2a37 |
327 | Make Perl buildable with a cross-compiler. This will play havoc with |
da75cd15 |
328 | Configure, which needs to know how the target system will respond to |
722d2a37 |
329 | its tests; maybe C<microperl> will be a good starting point here. |
330 | (Indeed, Bart Schuller reports that he compiled up C<microperl> for |
331 | the Agenda PDA and it works fine.) A really big spanner in the works |
332 | is the bootstrapping build process of Perl: if the filesystem the |
333 | target systems sees is not the same what the build host sees, various |
334 | input, output, and (Perl) library files need to be copied back and forth. |
e50bb9a1 |
335 | |
f86a8bc5 |
336 | As of 5.8.0 Configure mostly works for cross-compilation |
337 | (used successfully for iPAQ Linux), miniperl gets built, |
338 | but then building DynaLoader (and other extensions) fails |
339 | since MakeMaker knows nothing of cross-compilation. |
340 | (See INSTALL/Cross-compilation for the state of things.) |
341 | |
722d2a37 |
342 | =head2 Perl preprocessor / macros |
e50bb9a1 |
343 | |
722d2a37 |
344 | Source filters help with this, but do not get us all the way. For |
345 | instance, it should be possible to implement the C<??> operator somehow; |
346 | source filters don't (quite) cut it. |
e50bb9a1 |
347 | |
722d2a37 |
348 | =head2 Perl lexer in Perl |
a45bd81d |
349 | |
722d2a37 |
350 | Damian Conway is planning to work on this, but it hasn't happened yet. |
e50bb9a1 |
351 | |
722d2a37 |
352 | =head2 Using POSIX calls internally |
e50bb9a1 |
353 | |
210b36aa |
354 | When faced with a BSD vs. SysV -style interface to some library or |
722d2a37 |
355 | system function, perl's roots show in that it typically prefers the BSD |
356 | interface (but falls back to the SysV one). One example is getpgrp(). |
357 | Other examples include C<memcpy> vs. C<bcopy>. There are others, mostly in |
210b36aa |
358 | F<pp_sys.c>. |
e50bb9a1 |
359 | |
722d2a37 |
360 | Mostly, this item is a suggestion for which way to start a journey into |
361 | an C<#ifdef> forest. It is not primarily a suggestion to eliminate any of |
362 | the C<#ifdef> forests. |
e50bb9a1 |
363 | |
722d2a37 |
364 | POSIX calls are perhaps more likely to be portable to unexpected |
365 | architectures. They are also perhaps more likely to be actively |
366 | maintained by a current vendor. They are also perhaps more likely to be |
367 | available in thread-safe versions, if appropriate. |
e50bb9a1 |
368 | |
722d2a37 |
369 | =head2 -i rename file when changed |
e50bb9a1 |
370 | |
722d2a37 |
371 | It's only necessary to rename a file when inplace editing when the file |
372 | has changed. Detecting a change is perhaps the difficult bit. |
e50bb9a1 |
373 | |
722d2a37 |
374 | =head2 All ARGV input should act like E<lt>E<gt> |
e50bb9a1 |
375 | |
2d84a16a |
376 | eg C<read(ARGV, ...)> doesn't currently read across multiple files. |
377 | |
722d2a37 |
378 | =head2 Support for rerunning debugger |
e50bb9a1 |
379 | |
722d2a37 |
380 | There should be a way of restarting the debugger on demand. |
e50bb9a1 |
381 | |
c6287c21 |
382 | =head2 Test Suite for the Debugger |
383 | |
384 | The debugger is a complex piece of software and fixing something |
385 | here may inadvertently break something else over there. To tame |
386 | this chaotic behaviour, a test suite is necessary. |
387 | |
722d2a37 |
388 | =head2 my sub foo { } |
c47ff5f1 |
389 | |
722d2a37 |
390 | The basic principle is sound, but there are problems with the semantics |
391 | of self-referential and mutually referential lexical subs: how to |
392 | declare the subs? |
c47ff5f1 |
393 | |
722d2a37 |
394 | =head2 One-pass global destruction |
c47ff5f1 |
395 | |
722d2a37 |
396 | Sweeping away all the allocated memory in one go is a laudable goal, but |
397 | it's difficult and in most cases, it's easier to let the memory get |
398 | freed by exiting. |
e50bb9a1 |
399 | |
722d2a37 |
400 | =head2 Rewrite regexp parser |
e50bb9a1 |
401 | |
722d2a37 |
402 | There has been talk recently of rewriting the regular expression parser |
403 | to produce an optree instead of a chain of opcodes; it's unclear whether |
404 | or not this would be a win. |
e50bb9a1 |
405 | |
722d2a37 |
406 | =head2 Cache recently used regexps |
e50bb9a1 |
407 | |
722d2a37 |
408 | This is to speed up |
e50bb9a1 |
409 | |
722d2a37 |
410 | for my $re (@regexps) { |
411 | $matched++ if /$re/ |
412 | } |
e50bb9a1 |
413 | |
722d2a37 |
414 | C<qr//> already gives us a way of saving compiled regexps, but it should |
415 | be done automatically. |
e50bb9a1 |
416 | |
722d2a37 |
417 | =head2 Re-entrant functions |
e50bb9a1 |
418 | |
722d2a37 |
419 | Add configure probes for C<_r> forms of system calls and fit them to the |
420 | core. Unfortunately, calling conventions for these functions and not |
421 | standardised. |
04c70446 |
422 | |
722d2a37 |
423 | =head2 Cross-compilation support |
04c70446 |
424 | |
722d2a37 |
425 | Bart Schuller reports that using C<microperl> and a cross-compiler, he |
426 | got Perl working on the Agenda PDA. However, one cannot build a full |
427 | Perl because Configure needs to get the results for the target platform, |
428 | for the host. |
e50bb9a1 |
429 | |
722d2a37 |
430 | =head2 Bit-shifting bitvectors |
e50bb9a1 |
431 | |
722d2a37 |
432 | Given: |
e50bb9a1 |
433 | |
722d2a37 |
434 | vec($v, 1000, 1) = 1; |
e50bb9a1 |
435 | |
722d2a37 |
436 | One should be able to do |
e50bb9a1 |
437 | |
722d2a37 |
438 | $v <<= 1; |
e50bb9a1 |
439 | |
722d2a37 |
440 | and have the 999'th bit set. |
e50bb9a1 |
441 | |
722d2a37 |
442 | Currently if you try with shift bitvectors you shift the NV/UV, instead |
443 | of the bits in the PV. Not very logical. |
e50bb9a1 |
444 | |
722d2a37 |
445 | =head2 debugger pragma |
e50bb9a1 |
446 | |
722d2a37 |
447 | The debugger is implemented in Perl in F<perl5db.pl>; turning it into a |
448 | pragma should be easy, but making it work lexically might be more |
449 | difficult. Fiddling with C<$^P> would be necessary. |
e50bb9a1 |
450 | |
722d2a37 |
451 | =head2 use less pragma |
e50bb9a1 |
452 | |
722d2a37 |
453 | Identify areas where speed/memory tradeoffs can be made and have a hint |
454 | to switch between them. |
e50bb9a1 |
455 | |
722d2a37 |
456 | =head2 switch structures |
e50bb9a1 |
457 | |
722d2a37 |
458 | Although we have C<Switch.pm> in core, Larry points to the dormant |
459 | C<nswitch> and C<cswitch> ops in F<pp.c>; using these opcodes would be |
460 | much faster. |
e50bb9a1 |
461 | |
722d2a37 |
462 | =head2 Cache eval tree |
e50bb9a1 |
463 | |
722d2a37 |
464 | =head2 rcatmaybe |
e50bb9a1 |
465 | |
722d2a37 |
466 | =head2 Shrink opcode tables |
e50bb9a1 |
467 | |
722d2a37 |
468 | =head2 Optimize away @_ |
e50bb9a1 |
469 | |
722d2a37 |
470 | Look at the "reification" code in C<av.c> |
e50bb9a1 |
471 | |
722d2a37 |
472 | =head2 Prototypes versus indirect objects |
e50bb9a1 |
473 | |
722d2a37 |
474 | Currently, indirect object syntax bypasses prototype checks. |
e50bb9a1 |
475 | |
210b36aa |
476 | =head2 Install HTML |
e50bb9a1 |
477 | |
722d2a37 |
478 | HTML versions of the documentation need to be installed by default; a |
479 | call to C<installhtml> from C<installperl> may be all that's necessary. |
e50bb9a1 |
480 | |
722d2a37 |
481 | =head2 Prototype method calls |
e50bb9a1 |
482 | |
722d2a37 |
483 | =head2 Return context prototype declarations |
e50bb9a1 |
484 | |
722d2a37 |
485 | =head2 magic_setisa |
e50bb9a1 |
486 | |
722d2a37 |
487 | =head2 Garbage collection |
e50bb9a1 |
488 | |
722d2a37 |
489 | There have been persistent mumblings about putting a mark-and-sweep |
490 | garbage detector into Perl; Alan Burlison has some ideas about this. |
e50bb9a1 |
491 | |
722d2a37 |
492 | =head2 IO tutorial |
e50bb9a1 |
493 | |
722d2a37 |
494 | Mark-Jason Dominus has the beginnings of one of these. |
e50bb9a1 |
495 | |
722d2a37 |
496 | =head2 Rewrite perldoc |
e50bb9a1 |
497 | |
722d2a37 |
498 | There are a few suggestions for what to do with C<perldoc>: maybe a |
499 | full-text search, an index function, locating pages on a particular |
500 | high-level subject, and so on. |
e50bb9a1 |
501 | |
3958b146 |
502 | =head2 Install .3p manpages |
e50bb9a1 |
503 | |
3958b146 |
504 | This is a bone of contention; we can create C<.3p> manpages for each |
722d2a37 |
505 | built-in function, but should we install them by default? Tcl does this, |
506 | and it clutters up C<apropos>. |
e50bb9a1 |
507 | |
722d2a37 |
508 | =head2 Unicode tutorial |
e50bb9a1 |
509 | |
722d2a37 |
510 | Simon Cozens promises to do this before he gets old. |
e50bb9a1 |
511 | |
722d2a37 |
512 | =head2 Update POSIX.pm for 1003.1-2 |
3958b146 |
513 | |
722d2a37 |
514 | =head2 Retargetable installation |
e50bb9a1 |
515 | |
722d2a37 |
516 | Allow C<@INC> to be changed after Perl is built. |
e50bb9a1 |
517 | |
722d2a37 |
518 | =head2 POSIX emulation on non-POSIX systems |
e50bb9a1 |
519 | |
722d2a37 |
520 | Make C<POSIX.pm> behave as POSIXly as possible everywhere, meaning we |
521 | have to implement POSIX equivalents for some functions if necessary. |
e50bb9a1 |
522 | |
722d2a37 |
523 | =head2 Rename Win32 headers |
e50bb9a1 |
524 | |
722d2a37 |
525 | =head2 Finish off lvalue functions |
526 | |
527 | They don't work in the debugger, and they don't work for list or hash |
528 | slices. |
e50bb9a1 |
529 | |
722d2a37 |
530 | =head2 Update sprintf documentation |
e50bb9a1 |
531 | |
722d2a37 |
532 | Hugo van der Sanden plans to look at this. |
e50bb9a1 |
533 | |
722d2a37 |
534 | =head2 Use fchown/fchmod internally |
e50bb9a1 |
535 | |
722d2a37 |
536 | This has been done in places, but needs a thorough code review. |
537 | Also fchdir is available in some platforms. |
e50bb9a1 |
538 | |
d45541b3 |
539 | =head2 Make v-strings overloaded objects |
c5fc23ff |
540 | |
d45541b3 |
541 | Instead of having to guess whether a string is a v-string and thus |
542 | needs to be displayed with %vd, make v-strings (readonly) objects |
543 | (class "vstring"?) with a stringify overload. |
c5fc23ff |
544 | |
49293501 |
545 | =head2 Allow restricted hash assignment |
546 | |
547 | Currently you're not allowed to assign to a restricted hash at all, |
548 | even with the same keys. |
549 | |
550 | %restricted = (foo => 42); # error |
551 | |
552 | This should be allowed if the new keyset is a subset of the old |
553 | keyset. May require more extra code than we'd like in pp_aassign. |
554 | |
5387ccf1 |
555 | =head2 Should overload be inheritable? |
556 | |
557 | Should overload be 'contagious' through @ISA so that derived classes |
558 | would inherit their base classes' overload definitions? What to do |
559 | in case of overload conflicts? |
560 | |
cbda53d5 |
561 | =head2 Taint rethink |
562 | |
563 | Should taint be stopped from affecting control flow, if ($tainted)? |
564 | Should tainted symbolic method calls and subref calls be stopped? |
565 | (Look at Ruby's $SAFE levels for inspiration?) |
566 | |
722d2a37 |
567 | =head1 Vague ideas |
e50bb9a1 |
568 | |
722d2a37 |
569 | Ideas which have been discussed, and which may or may not happen. |
e50bb9a1 |
570 | |
722d2a37 |
571 | =head2 ref() in list context |
e50bb9a1 |
572 | |
722d2a37 |
573 | It's unclear what this should do or how to do it without breaking old |
574 | code. |
e50bb9a1 |
575 | |
f86a8bc5 |
576 | =head2 Make tr/// return histogram of characters in list context |
e50bb9a1 |
577 | |
722d2a37 |
578 | There is a patch for this, but it may require Unicodification. |
e50bb9a1 |
579 | |
722d2a37 |
580 | =head2 Compile to real threaded code |
3958b146 |
581 | |
722d2a37 |
582 | =head2 Structured types |
3958b146 |
583 | |
722d2a37 |
584 | =head2 Modifiable $1 et al. |
e50bb9a1 |
585 | |
722d2a37 |
586 | ($x = "elephant") =~ /e(ph)/; |
587 | $1 = "g"; # $x = "elegant" |
e50bb9a1 |
588 | |
722d2a37 |
589 | What happens if there are multiple (nested?) brackets? What if the |
590 | string changes between the match and the assignment? |
e50bb9a1 |
591 | |
722d2a37 |
592 | =head2 Procedural interfaces for IO::*, etc. |
e50bb9a1 |
593 | |
722d2a37 |
594 | Some core modules have been accused of being overly-OO. Adding |
595 | procedural interfaces could demystify them. |
e50bb9a1 |
596 | |
722d2a37 |
597 | =head2 RPC modules |
e50bb9a1 |
598 | |
722d2a37 |
599 | =head2 Attach/detach debugger from running program |
e50bb9a1 |
600 | |
722d2a37 |
601 | With C<gdb>, you can attach the debugger to a running program if you |
602 | pass the process ID. It would be good to do this with the Perl debugger |
603 | on a running Perl program, although I'm not sure how it would be done. |
e50bb9a1 |
604 | |
722d2a37 |
605 | =head2 GUI::Native |
e50bb9a1 |
606 | |
722d2a37 |
607 | A non-core module that would use "native" GUI to create graphical |
608 | applications. |
e50bb9a1 |
609 | |
722d2a37 |
610 | =head2 foreach(reverse ...) |
e50bb9a1 |
611 | |
722d2a37 |
612 | Currently |
e50bb9a1 |
613 | |
722d2a37 |
614 | foreach (reverse @_) { ... } |
e50bb9a1 |
615 | |
722d2a37 |
616 | puts C<@_> on the stack, reverses it putting the reversed version on the |
617 | stack, then iterates forwards. Instead, it could be special-cased to put |
618 | C<@_> on the stack then iterate backwards. |
e50bb9a1 |
619 | |
722d2a37 |
620 | =head2 Constant function cache |
e50bb9a1 |
621 | |
722d2a37 |
622 | =head2 Approximate regular expression matching |
e50bb9a1 |
623 | |
722d2a37 |
624 | =head1 Ongoing |
e50bb9a1 |
625 | |
722d2a37 |
626 | These items B<always> need doing: |
e50bb9a1 |
627 | |
722d2a37 |
628 | =head2 Update guts documentation |
e50bb9a1 |
629 | |
722d2a37 |
630 | Simon Cozens tries to do this when possible, and contributions to the |
631 | C<perlapi> documentation is welcome. |
e50bb9a1 |
632 | |
722d2a37 |
633 | =head2 Add more tests |
e50bb9a1 |
634 | |
722d2a37 |
635 | Michael Schwern will donate $500 to Yet Another Society when all core |
636 | modules have tests. |
e50bb9a1 |
637 | |
722d2a37 |
638 | =head2 Update auxiliary tools |
e50bb9a1 |
639 | |
722d2a37 |
640 | The code we ship with Perl should look like good Perl 5. |
e50bb9a1 |
641 | |
1e278fd9 |
642 | =head2 Create debugging macros |
643 | |
644 | Debugging macros (like printsv, dump) can make debugging perl inside a |
645 | C debugger much easier. A good set for gdb comes with mod_perl. |
646 | Something similar should be distributed with perl. |
647 | |
648 | The proper way to do this is to use and extend Devel::DebugInit. |
649 | Devel::DebugInit also needs to be extended to support threads. |
650 | |
651 | See p5p archives for late May/early June 2001 for a recent discussion |
652 | on this topic. |
653 | |
654 | =head2 truncate to the people |
655 | |
656 | One can emulate ftruncate() using F_FREESP and F_CHSIZ fcntls |
657 | (see the UNIX FAQ for details). This needs to go somewhere near |
658 | pp_sys.c:pp_truncate(). |
659 | |
660 | One can emulate truncate() easily if one has ftruncate(). |
661 | This emulation should also go near pp_sys.pp_truncate(). |
662 | |
663 | =head2 Unicode in Filenames |
664 | |
665 | chdir, chmod, chown, chroot, exec, glob, link, lstat, mkdir, open, qx, |
666 | readdir, readlink, rename, rmdir, stat, symlink, sysopen, system, |
667 | truncate, unlink, utime. All these could potentially accept Unicode |
668 | filenames either as input or output (and in the case of system and qx |
669 | Unicode in general, as input or output to/from the shell). Whether a |
670 | filesystem - an operating system pair understands Unicode in filenames |
671 | varies. |
672 | |
673 | Known combinations that have some level of understanding include |
674 | Microsoft NTFS, Apple HFS+ (In Mac OS 9 and X) and Apple UFS (in Mac |
675 | OS X), NFS v4 is rumored to be Unicode, and of course Plan 9. How to |
676 | create Unicode filenames, what forms of Unicode are accepted and used |
677 | (UCS-2, UTF-16, UTF-8), what (if any) is the normalization form used, |
678 | and so on, varies. Finding the right level of interfacing to Perl |
679 | requires some thought. Remember that an OS does not implicate a |
680 | filesystem. |
681 | |
eb450546 |
682 | Note that in Windows the -C command line flag already does quite |
683 | a bit of the above (but even there the support is not complete: |
684 | for example the exec/spawn are not Unicode-aware) by turning on |
685 | the so-called "wide API support". |
686 | |
722d2a37 |
687 | =head1 Recently done things |
e50bb9a1 |
688 | |
722d2a37 |
689 | These are things which have been on the todo lists in previous releases |
690 | but have recently been completed. |
e50bb9a1 |
691 | |
b0b7f283 |
692 | =head2 Alternative RE syntax module |
693 | |
694 | The C<Regexp::English> module, available from the CPAN, provides this: |
695 | |
696 | my $re = Regexp::English |
697 | -> start_of_line |
698 | -> literal('Flippers') |
699 | -> literal(':') |
700 | -> optional |
701 | -> whitespace_char |
702 | -> end |
703 | -> remember |
704 | -> multiple |
705 | -> digit; |
706 | |
707 | /$re/; |
708 | |
722d2a37 |
709 | =head2 Safe signal handling |
e50bb9a1 |
710 | |
722d2a37 |
711 | A new signal model went into 5.7.1 without much fanfare. Operations and |
712 | C<malloc>s are no longer interrupted by signals, which are handled |
713 | between opcodes. This means that C<PERL_ASYNC_CHECK> now actually does |
714 | something. However, there are still a few things that need to be done. |
e50bb9a1 |
715 | |
722d2a37 |
716 | =head2 Tie Modules |
e50bb9a1 |
717 | |
722d2a37 |
718 | Modules which implement arrays in terms of strings, substrings or files |
719 | can be found on the CPAN. |
e50bb9a1 |
720 | |
722d2a37 |
721 | =head2 gettimeofday |
e50bb9a1 |
722 | |
210b36aa |
723 | C<Time::HiRes> has been integrated into the core. |
e50bb9a1 |
724 | |
722d2a37 |
725 | =head2 setitimer and getimiter |
e50bb9a1 |
726 | |
210b36aa |
727 | Adding C<Time::HiRes> got us this too. |
e50bb9a1 |
728 | |
722d2a37 |
729 | =head2 Testing __DIE__ hook |
730 | |
731 | Tests have been added. |
732 | |
733 | =head2 CPP equivalent in Perl |
e50bb9a1 |
734 | |
722d2a37 |
735 | A C Yardley will probably have done this by the time you can read this. |
736 | This allows for a generalization of the C constant detection used in |
737 | building C<Errno.pm>. |
e50bb9a1 |
738 | |
722d2a37 |
739 | =head2 Explicit switch statements |
e50bb9a1 |
740 | |
722d2a37 |
741 | C<Switch.pm> has been integrated into the core to give you all manner of |
742 | C<switch...case> semantics. |
e50bb9a1 |
743 | |
722d2a37 |
744 | =head2 autocroak |
e50bb9a1 |
745 | |
722d2a37 |
746 | This is C<Fatal.pm>. |
e50bb9a1 |
747 | |
722d2a37 |
748 | =head2 UTF/EBCDIC |
e50bb9a1 |
749 | |
722d2a37 |
750 | Nick Ing-Simmons has made UTF-EBCDIC (UTR13) work with Perl. |
e50bb9a1 |
751 | |
722d2a37 |
752 | EBCDIC? http://www.unicode.org/unicode/reports/tr16/ |
e50bb9a1 |
753 | |
722d2a37 |
754 | =head2 UTF Regexes |
e50bb9a1 |
755 | |
722d2a37 |
756 | Although there are probably some small bugs to be rooted out, Jarkko |
757 | Hietaniemi has made regular expressions polymorphic between bytes and |
758 | characters. |
e50bb9a1 |
759 | |
722d2a37 |
760 | =head2 perlcc to produce executable |
e50bb9a1 |
761 | |
722d2a37 |
762 | C<perlcc> was recently rewritten, and can now produce standalone |
763 | executables. |
e50bb9a1 |
764 | |
722d2a37 |
765 | =head2 END blocks saved in compiled output |
e50bb9a1 |
766 | |
722d2a37 |
767 | =head2 Secure temporary file module |
e50bb9a1 |
768 | |
722d2a37 |
769 | Tim Jenness' C<File::Temp> is now in core. |
e50bb9a1 |
770 | |
722d2a37 |
771 | =head2 Integrate Time::HiRes |
e50bb9a1 |
772 | |
722d2a37 |
773 | This module is now part of core. |
e50bb9a1 |
774 | |
722d2a37 |
775 | =head2 Turn Cwd into XS |
e50bb9a1 |
776 | |
722d2a37 |
777 | Benjamin Sugars has done this. |
e50bb9a1 |
778 | |
722d2a37 |
779 | =head2 Mmap for input |
e50bb9a1 |
780 | |
722d2a37 |
781 | Nick Ing-Simmons' C<perlio> supports an C<mmap> IO method. |
e50bb9a1 |
782 | |
722d2a37 |
783 | =head2 Byte to/from UTF8 and UTF8 to/from local conversion |
e50bb9a1 |
784 | |
722d2a37 |
785 | C<Encode> provides this. |
e50bb9a1 |
786 | |
722d2a37 |
787 | =head2 Add sockatmark support |
e50bb9a1 |
788 | |
722d2a37 |
789 | Added in 5.7.1 |
e50bb9a1 |
790 | |
722d2a37 |
791 | =head2 Mailing list archives |
792 | |
f224927c |
793 | http://lists.perl.org/ , http://archive.develooper.com/ |
722d2a37 |
794 | |
795 | =head2 Bug tracking |
796 | |
797 | Richard Foley has written the bug tracking system at http://bugs.perl.org/ |
e50bb9a1 |
798 | |
722d2a37 |
799 | =head2 Integrate MacPerl |
e50bb9a1 |
800 | |
722d2a37 |
801 | Chris Nandor and Matthias Neeracher have integrated the MacPerl changes |
802 | into 5.6.0. |
e50bb9a1 |
803 | |
722d2a37 |
804 | =head2 Web "nerve center" for Perl |
e50bb9a1 |
805 | |
722d2a37 |
806 | http://use.perl.org/ is what you're looking for. |
e50bb9a1 |
807 | |
722d2a37 |
808 | =head2 Regular expression tutorial |
e50bb9a1 |
809 | |
722d2a37 |
810 | C<perlretut>, provided by Mark Kvale. |
e50bb9a1 |
811 | |
722d2a37 |
812 | =head2 Debugging Tutorial |
e50bb9a1 |
813 | |
722d2a37 |
814 | C<perldebtut>, written by Richard Foley. |
e50bb9a1 |
815 | |
722d2a37 |
816 | =head2 Integrate new modules |
e50bb9a1 |
817 | |
722d2a37 |
818 | Jarkko has been integrating madly into 5.7.x |
e50bb9a1 |
819 | |
722d2a37 |
820 | =head2 Integrate profiler |
e50bb9a1 |
821 | |
722d2a37 |
822 | C<Devel::DProf> is now a core module. |
e50bb9a1 |
823 | |
722d2a37 |
824 | =head2 Y2K error detection |
e50bb9a1 |
825 | |
722d2a37 |
826 | There's a configure option to detect unsafe concatenation with "19", and |
827 | a CPAN module. (C<D'oh::Year>) |
e50bb9a1 |
828 | |
722d2a37 |
829 | =head2 Regular expression debugger |
e50bb9a1 |
830 | |
722d2a37 |
831 | While not part of core, Mark-Jason Dominus has written C<Rx> and has |
832 | also come up with a generalised strategy for regular expression |
833 | debugging. |
e50bb9a1 |
834 | |
722d2a37 |
835 | =head2 POD checker |
e50bb9a1 |
836 | |
722d2a37 |
837 | That's, uh, F<podchecker> |
e50bb9a1 |
838 | |
722d2a37 |
839 | =head2 "Dynamic" lexicals |
e50bb9a1 |
840 | |
722d2a37 |
841 | =head2 Cache precompiled modules |
e50bb9a1 |
842 | |
722d2a37 |
843 | =head1 Deprecated Wishes |
e50bb9a1 |
844 | |
722d2a37 |
845 | These are items which used to be in the todo file, but have been |
846 | deprecated for some reason. |
e50bb9a1 |
847 | |
722d2a37 |
848 | =head2 Loop control on do{} |
e50bb9a1 |
849 | |
722d2a37 |
850 | This would break old code; use C<do{{ }}> instead. |
e50bb9a1 |
851 | |
722d2a37 |
852 | =head2 Lexically scoped typeglobs |
e50bb9a1 |
853 | |
722d2a37 |
854 | Not needed now we have lexical IO handles. |
e50bb9a1 |
855 | |
722d2a37 |
856 | =head2 format BOTTOM |
3958b146 |
857 | |
722d2a37 |
858 | =head2 report HANDLE |
e50bb9a1 |
859 | |
722d2a37 |
860 | Damian Conway's text formatting modules seem to be the Way To Go. |
e50bb9a1 |
861 | |
722d2a37 |
862 | =head2 Generalised want()/caller()) |
3958b146 |
863 | |
638ae6a9 |
864 | Robin Houston's C<Want> module does this. |
865 | |
722d2a37 |
866 | =head2 Named prototypes |
e50bb9a1 |
867 | |
638ae6a9 |
868 | This seems to be delayed until Perl 6. |
e50bb9a1 |
869 | |
722d2a37 |
870 | =head2 Built-in globbing |
e50bb9a1 |
871 | |
722d2a37 |
872 | The C<File::Glob> module has been used to replace the C<glob> function. |
e50bb9a1 |
873 | |
722d2a37 |
874 | =head2 Regression tests for suidperl |
e50bb9a1 |
875 | |
722d2a37 |
876 | C<suidperl> is deprecated in favour of common sense. |
e50bb9a1 |
877 | |
722d2a37 |
878 | =head2 Cached hash values |
e50bb9a1 |
879 | |
722d2a37 |
880 | We have shared hash keys, which perform the same job. |
e50bb9a1 |
881 | |
722d2a37 |
882 | =head2 Add compression modules |
e50bb9a1 |
883 | |
722d2a37 |
884 | The compression modules are a little heavy; meanwhile, Nick Clark is |
885 | working on experimental pragmata to do transparent decompression on |
886 | input. |
e50bb9a1 |
887 | |
722d2a37 |
888 | =head2 Reorganise documentation into tutorials/references |
e50bb9a1 |
889 | |
722d2a37 |
890 | Could not get consensus on P5P about this. |
e50bb9a1 |
891 | |
722d2a37 |
892 | =head2 Remove distinction between functions and operators |
893 | |
894 | Caution: highly flammable. |
895 | |
896 | =head2 Make XS easier to use |
e50bb9a1 |
897 | |
722d2a37 |
898 | Use C<Inline> instead, or SWIG. |
e50bb9a1 |
899 | |
722d2a37 |
900 | =head2 Make embedding easier to use |
e50bb9a1 |
901 | |
722d2a37 |
902 | Use C<Inline::CPR>. |
e50bb9a1 |
903 | |
722d2a37 |
904 | =head2 man for perl |
04c70446 |
905 | |
1577cd80 |
906 | See the Perl Power Tools. ( http://language.perl.com/ppt/ ) |
04c70446 |
907 | |
722d2a37 |
908 | =head2 my $Package::variable |
04c70446 |
909 | |
722d2a37 |
910 | Use C<our> instead. |
04c70446 |
911 | |
722d2a37 |
912 | =head2 "or" tests defined, not truth |
04c70446 |
913 | |
722d2a37 |
914 | Suggesting this on P5P B<will> cause a boring and interminable flamewar. |
04c70446 |
915 | |
722d2a37 |
916 | =head2 "class"-based lexicals |
04c70446 |
917 | |
cbb3fa72 |
918 | Use flyweight objects, secure hashes or, dare I say it, pseudo-hashes instead. |
f86a8bc5 |
919 | (Or whatever will replace pseudohashes in 5.10.) |
04c70446 |
920 | |
722d2a37 |
921 | =head2 byteperl |
04c70446 |
922 | |
722d2a37 |
923 | C<ByteLoader> covers this. |
04c70446 |
924 | |
722d2a37 |
925 | =head2 Lazy evaluation / tail recursion removal |
04c70446 |
926 | |
f86a8bc5 |
927 | C<List::Util> gives first() (a short-circuiting grep); tail recursion |
928 | removal is done manually, with C<goto &whoami;>. (However, MJD has |
929 | found that C<goto &whoami> introduces a performance penalty, so maybe |
930 | there should be a way to do this after all: C<sub foo {START: ... goto |
931 | START;> is better.) |
0562c0e3 |
932 | |
933 | =head2 Make "use utf8" the default |
934 | |
f86a8bc5 |
935 | Because of backward compatibility this is difficult: scripts could not |
936 | contain B<any legacy eight-bit data> (like Latin-1) anymore, even in |
937 | string literals or pod. Also would introduce a measurable slowdown of |
938 | at least few percentages since all regular expression operations would |
939 | be done in full UTF-8. But if you want to try this, add |
940 | -DUSE_UTF8_SCRIPTS to your compilation flags. |
941 | |
3298bd4d |
942 | =head2 Unicode collation and normalization |
943 | |
944 | The Unicode::Collate and Unicode::Normalize modules |
945 | by SADAHIRO Tomoyuki have been included since 5.8.0. |
946 | |
947 | Collation? http://www.unicode.org/unicode/reports/tr10/ |
948 | Normalization? http://www.unicode.org/unicode/reports/tr15/ |
0562c0e3 |
949 | |
1626a787 |
950 | =head2 pack/unpack tutorial |
951 | |
952 | Wolfgang Laun finished what Simon Cozens started. |
953 | |
3298bd4d |
954 | =cut |