Commit | Line | Data |
---|---|---|
e41182b5 GS |
1 | =head1 NAME |
2 | ||
3 | perlport - Writing portable Perl | |
4 | ||
e41182b5 GS |
5 | =head1 DESCRIPTION |
6 | ||
b7df3edc GS |
7 | Perl runs on numerous operating systems. While most of them share |
8 | much in common, they also have their own unique features. | |
e41182b5 GS |
9 | |
10 | This document is meant to help you to find out what constitutes portable | |
b7df3edc | 11 | Perl code. That way once you make a decision to write portably, |
e41182b5 GS |
12 | you know where the lines are drawn, and you can stay within them. |
13 | ||
b7df3edc GS |
14 | There is a tradeoff between taking full advantage of one particular |
15 | type of computer and taking advantage of a full range of them. | |
16 | Naturally, as you broaden your range and become more diverse, the | |
17 | common factors drop, and you are left with an increasingly smaller | |
18 | area of common ground in which you can operate to accomplish a | |
19 | particular task. Thus, when you begin attacking a problem, it is | |
20 | important to consider under which part of the tradeoff curve you | |
21 | want to operate. Specifically, you must decide whether it is | |
2c044526 | 22 | important that the task that you are coding has the full generality |
b7df3edc GS |
23 | of being portable, or whether to just get the job done right now. |
24 | This is the hardest choice to be made. The rest is easy, because | |
25 | Perl provides many choices, whichever way you want to approach your | |
0a47030a GS |
26 | problem. |
27 | ||
28 | Looking at it another way, writing portable code is usually about | |
b7df3edc GS |
29 | willfully limiting your available choices. Naturally, it takes |
30 | discipline and sacrifice to do that. The product of portability | |
31 | and convenience may be a constant. You have been warned. | |
e41182b5 GS |
32 | |
33 | Be aware of two important points: | |
34 | ||
35 | =over 4 | |
36 | ||
37 | =item Not all Perl programs have to be portable | |
38 | ||
b7df3edc | 39 | There is no reason you should not use Perl as a language to glue Unix |
e41182b5 GS |
40 | tools together, or to prototype a Macintosh application, or to manage the |
41 | Windows registry. If it makes no sense to aim for portability for one | |
42 | reason or another in a given program, then don't bother. | |
43 | ||
b7df3edc | 44 | =item Nearly all of Perl already I<is> portable |
e41182b5 GS |
45 | |
46 | Don't be fooled into thinking that it is hard to create portable Perl | |
47 | code. It isn't. Perl tries its level-best to bridge the gaps between | |
48 | what's available on different platforms, and all the means available to | |
49 | use those features. Thus almost all Perl code runs on any machine | |
6ab3f9cb | 50 | without modification. But there are some significant issues in |
e41182b5 GS |
51 | writing portable code, and this document is entirely about those issues. |
52 | ||
53 | =back | |
54 | ||
b7df3edc GS |
55 | Here's the general rule: When you approach a task commonly done |
56 | using a whole range of platforms, think about writing portable | |
e41182b5 GS |
57 | code. That way, you don't sacrifice much by way of the implementation |
58 | choices you can avail yourself of, and at the same time you can give | |
59 | your users lots of platform choices. On the other hand, when you have to | |
60 | take advantage of some unique feature of a particular platform, as is | |
61 | often the case with systems programming (whether for Unix, Windows, | |
204ad8d5 | 62 | VMS, etc.), consider writing platform-specific code. |
e41182b5 | 63 | |
b7df3edc GS |
64 | When the code will run on only two or three operating systems, you |
65 | may need to consider only the differences of those particular systems. | |
66 | The important thing is to decide where the code will run and to be | |
0a47030a GS |
67 | deliberate in your decision. |
68 | ||
69 | The material below is separated into three main sections: main issues of | |
5a0de581 | 70 | portability (L</"ISSUES">), platform-specific issues (L</"PLATFORMS">), and |
2c044526 | 71 | built-in Perl functions that behave differently on various ports |
5a0de581 | 72 | (L</"FUNCTION IMPLEMENTATIONS">). |
e41182b5 GS |
73 | |
74 | This information should not be considered complete; it includes possibly | |
b8099c3d | 75 | transient information about idiosyncrasies of some of the ports, almost |
b7df3edc | 76 | all of which are in a state of constant evolution. Thus, this material |
e41182b5 | 77 | should be considered a perpetual work in progress |
cc07ed0b | 78 | (C<< <IMG SRC="yellow_sign.gif" ALT="Under Construction"> >>). |
e41182b5 | 79 | |
e41182b5 GS |
80 | =head1 ISSUES |
81 | ||
82 | =head2 Newlines | |
83 | ||
638bc118 | 84 | In most operating systems, lines in files are terminated by newlines. |
e41182b5 | 85 | Just what is used as a newline may vary from OS to OS. Unix |
b7df3edc | 86 | traditionally uses C<\012>, one type of DOSish I/O uses C<\015\012>, |
2c044526 | 87 | S<Mac OS> uses C<\015>, and z/OS uses C<\025>. |
e41182b5 | 88 | |
b7df3edc GS |
89 | Perl uses C<\n> to represent the "logical" newline, where what is |
90 | logical may depend on the platform in use. In MacPerl, C<\n> always | |
eb9df707 KW |
91 | means C<\015>. On EBCDIC platforms, C<\n> could be C<\025> or C<\045>. |
92 | In DOSish perls, C<\n> usually means C<\012>, but when | |
51d9476f | 93 | accessing a file in "text" mode, perl uses the C<:crlf> layer that |
94 | translates it to (or from) C<\015\012>, depending on whether you're | |
95 | reading or writing. Unix does the same thing on ttys in canonical | |
96 | mode. C<\015\012> is commonly referred to as CRLF. | |
b7df3edc | 97 | |
83a46a63 LM |
98 | To trim trailing newlines from text lines use |
99 | L<C<chomp>|perlfunc/chomp VARIABLE>. With default settings that function | |
100 | looks for a trailing C<\n> character and thus trims in a portable way. | |
5b3eff12 MS |
101 | |
102 | When dealing with binary files (or text files in binary mode) be sure | |
83a46a63 LM |
103 | to explicitly set L<C<$E<sol>>|perlvar/$E<sol>> to the appropriate value for |
104 | your file format before using L<C<chomp>|perlfunc/chomp VARIABLE>. | |
105 | ||
106 | Because of the "text" mode translation, DOSish perls have limitations in | |
107 | using L<C<seek>|perlfunc/seek FILEHANDLE,POSITION,WHENCE> and | |
108 | L<C<tell>|perlfunc/tell FILEHANDLE> on a file accessed in "text" mode. | |
109 | Stick to L<C<seek>|perlfunc/seek FILEHANDLE,POSITION,WHENCE>-ing to | |
110 | locations you got from L<C<tell>|perlfunc/tell FILEHANDLE> (and no | |
111 | others), and you are usually free to use | |
112 | L<C<seek>|perlfunc/seek FILEHANDLE,POSITION,WHENCE> and | |
113 | L<C<tell>|perlfunc/tell FILEHANDLE> even in "text" mode. Using | |
114 | L<C<seek>|perlfunc/seek FILEHANDLE,POSITION,WHENCE> or | |
115 | L<C<tell>|perlfunc/tell FILEHANDLE> or other file operations may be | |
116 | non-portable. If you use L<C<binmode>|perlfunc/binmode FILEHANDLE> on a | |
117 | file, however, you can usually | |
118 | L<C<seek>|perlfunc/seek FILEHANDLE,POSITION,WHENCE> and | |
119 | L<C<tell>|perlfunc/tell FILEHANDLE> with arbitrary values safely. | |
e41182b5 | 120 | |
2c044526 | 121 | A common misconception in socket programming is that S<C<\n eq \012>> |
0a47030a | 122 | everywhere. When using protocols such as common Internet protocols, |
e41182b5 GS |
123 | C<\012> and C<\015> are called for specifically, and the values of |
124 | the logical C<\n> and C<\r> (carriage return) are not reliable. | |
125 | ||
83a46a63 LM |
126 | print $socket "Hi there, client!\r\n"; # WRONG |
127 | print $socket "Hi there, client!\015\012"; # RIGHT | |
e41182b5 | 128 | |
0a47030a GS |
129 | However, using C<\015\012> (or C<\cM\cJ>, or C<\x0D\x0A>) can be tedious |
130 | and unsightly, as well as confusing to those maintaining the code. As | |
83a46a63 LM |
131 | such, the L<C<Socket>|Socket> module supplies the Right Thing for those |
132 | who want it. | |
e41182b5 GS |
133 | |
134 | use Socket qw(:DEFAULT :crlf); | |
83a46a63 | 135 | print $socket "Hi there, client!$CRLF" # RIGHT |
e41182b5 | 136 | |
6ab3f9cb | 137 | When reading from a socket, remember that the default input record |
83a46a63 LM |
138 | separator L<C<$E<sol>>|perlvar/$E<sol>> is C<\n>, but robust socket code |
139 | will recognize as either C<\012> or C<\015\012> as end of line: | |
e41182b5 | 140 | |
83a46a63 | 141 | while (<$socket>) { # NOT ADVISABLE! |
e41182b5 GS |
142 | # ... |
143 | } | |
144 | ||
b7df3edc GS |
145 | Because both CRLF and LF end in LF, the input record separator can |
146 | be set to LF and any CR stripped later. Better to write: | |
e41182b5 GS |
147 | |
148 | use Socket qw(:DEFAULT :crlf); | |
149 | local($/) = LF; # not needed if $/ is already \012 | |
150 | ||
83a46a63 | 151 | while (<$socket>) { |
e41182b5 GS |
152 | s/$CR?$LF/\n/; # not sure if socket uses LF or CRLF, OK |
153 | # s/\015?\012/\n/; # same thing | |
154 | } | |
155 | ||
b7df3edc GS |
156 | This example is preferred over the previous one--even for Unix |
157 | platforms--because now any C<\015>'s (C<\cM>'s) are stripped out | |
e41182b5 GS |
158 | (and there was much rejoicing). |
159 | ||
6ab3f9cb | 160 | Similarly, functions that return text data--such as a function that |
b7df3edc GS |
161 | fetches a web page--should sometimes translate newlines before |
162 | returning the data, if they've not yet been translated to the local | |
163 | newline representation. A single line of code will often suffice: | |
2ee0eb3c | 164 | |
b7df3edc GS |
165 | $data =~ s/\015?\012/\n/g; |
166 | return $data; | |
2ee0eb3c | 167 | |
6ab3f9cb GS |
168 | Some of this may be confusing. Here's a handy reference to the ASCII CR |
169 | and LF characters. You can print it out and stick it in your wallet. | |
170 | ||
74555b7a PP |
171 | LF eq \012 eq \x0A eq \cJ eq chr(10) eq ASCII 10 |
172 | CR eq \015 eq \x0D eq \cM eq chr(13) eq ASCII 13 | |
6ab3f9cb GS |
173 | |
174 | | Unix | DOS | Mac | | |
175 | --------------------------- | |
176 | \n | LF | LF | CR | | |
177 | \r | CR | CR | LF | | |
178 | \n * | LF | CRLF | CR | | |
179 | \r * | CR | CR | LF | | |
180 | --------------------------- | |
181 | * text-mode STDIO | |
182 | ||
b7df3edc GS |
183 | The Unix column assumes that you are not accessing a serial line |
184 | (like a tty) in canonical mode. If you are, then CR on input becomes | |
185 | "\n", and "\n" on output becomes CRLF. | |
186 | ||
6ab3f9cb | 187 | These are just the most common definitions of C<\n> and C<\r> in Perl. |
522b859a JH |
188 | There may well be others. For example, on an EBCDIC implementation |
189 | such as z/OS (OS/390) or OS/400 (using the ILE, the PASE is ASCII-based) | |
190 | the above material is similar to "Unix" but the code numbers change: | |
74555b7a | 191 | |
d770bc45 TS |
192 | LF eq \025 eq \x15 eq \cU eq chr(21) eq CP-1047 21 |
193 | LF eq \045 eq \x25 eq chr(37) eq CP-0037 37 | |
74555b7a PP |
194 | CR eq \015 eq \x0D eq \cM eq chr(13) eq CP-1047 13 |
195 | CR eq \015 eq \x0D eq \cM eq chr(13) eq CP-0037 13 | |
196 | ||
197 | | z/OS | OS/400 | | |
198 | ---------------------- | |
199 | \n | LF | LF | | |
200 | \r | CR | CR | | |
201 | \n * | LF | LF | | |
202 | \r * | CR | CR | | |
203 | ---------------------- | |
204 | * text-mode STDIO | |
6ab3f9cb | 205 | |
322422de GS |
206 | =head2 Numbers endianness and Width |
207 | ||
208 | Different CPUs store integers and floating point numbers in different | |
209 | orders (called I<endianness>) and widths (32-bit and 64-bit being the | |
b7df3edc GS |
210 | most common today). This affects your programs when they attempt to transfer |
211 | numbers in binary format from one CPU architecture to another, | |
212 | usually either "live" via network connection, or by storing the | |
213 | numbers to secondary storage such as a disk file or tape. | |
322422de | 214 | |
2c044526 | 215 | Conflicting storage orders make an utter mess out of the numbers. If a |
d1e3b762 | 216 | little-endian host (Intel, VAX) stores 0x12345678 (305419896 in |
b84d4f81 JH |
217 | decimal), a big-endian host (Motorola, Sparc, PA) reads it as |
218 | 0x78563412 (2018915346 in decimal). Alpha and MIPS can be either: | |
219 | Digital/Compaq used/uses them in little-endian mode; SGI/Cray uses | |
220 | them in big-endian mode. To avoid this problem in network (socket) | |
83a46a63 LM |
221 | connections use the L<C<pack>|perlfunc/pack TEMPLATE,LIST> and |
222 | L<C<unpack>|perlfunc/unpack TEMPLATE,EXPR> formats C<n> and C<N>, the | |
b84d4f81 | 223 | "network" orders. These are guaranteed to be portable. |
322422de | 224 | |
2c044526 | 225 | As of Perl 5.10.0, you can also use the C<E<gt>> and C<E<lt>> modifiers |
1109a392 MHM |
226 | to force big- or little-endian byte-order. This is useful if you want |
227 | to store signed integers or 64-bit integers, for example. | |
228 | ||
d1e3b762 GS |
229 | You can explore the endianness of your platform by unpacking a |
230 | data structure packed in native format such as: | |
231 | ||
232 | print unpack("h*", pack("s2", 1, 2)), "\n"; | |
233 | # '10002000' on e.g. Intel x86 or Alpha 21064 in little-endian mode | |
234 | # '00100020' on e.g. Motorola 68040 | |
235 | ||
236 | If you need to distinguish between endian architectures you could use | |
237 | either of the variables set like so: | |
238 | ||
239 | $is_big_endian = unpack("h*", pack("s", 1)) =~ /01/; | |
4375e838 | 240 | $is_little_endian = unpack("h*", pack("s", 1)) =~ /^1/; |
d1e3b762 | 241 | |
b7df3edc GS |
242 | Differing widths can cause truncation even between platforms of equal |
243 | endianness. The platform of shorter width loses the upper parts of the | |
322422de GS |
244 | number. There is no good solution for this problem except to avoid |
245 | transferring or storing raw binary numbers. | |
246 | ||
b7df3edc | 247 | One can circumnavigate both these problems in two ways. Either |
322422de | 248 | transfer and store numbers always in text format, instead of raw |
83a46a63 LM |
249 | binary, or else consider using modules like |
250 | L<C<Data::Dumper>|Data::Dumper> and L<C<Storable>|Storable> (included as | |
251 | of Perl 5.8). Keeping all data as text significantly simplifies matters. | |
322422de | 252 | |
433acd8a | 253 | =head2 Files and Filesystems |
e41182b5 GS |
254 | |
255 | Most platforms these days structure files in a hierarchical fashion. | |
b7df3edc | 256 | So, it is reasonably safe to assume that all platforms support the |
6ab3f9cb | 257 | notion of a "path" to uniquely identify a file on the system. How |
b7df3edc | 258 | that path is really written, though, differs considerably. |
e41182b5 | 259 | |
4375e838 | 260 | Although similar, file path specifications differ between Unix, |
b7df3edc GS |
261 | Windows, S<Mac OS>, OS/2, VMS, VOS, S<RISC OS>, and probably others. |
262 | Unix, for example, is one of the few OSes that has the elegant idea | |
263 | of a single root directory. | |
322422de | 264 | |
6ab3f9cb GS |
265 | DOS, OS/2, VMS, VOS, and Windows can work similarly to Unix with C</> |
266 | as path separator, or in their own idiosyncratic ways (such as having | |
267 | several root directories and various "unrooted" device files such NIL: | |
268 | and LPT:). | |
322422de | 269 | |
204ad8d5 | 270 | S<Mac OS> 9 and earlier used C<:> as a path separator instead of C</>. |
322422de | 271 | |
83a46a63 LM |
272 | The filesystem may support neither hard links |
273 | (L<C<link>|perlfunc/link OLDFILE,NEWFILE>) nor symbolic links | |
274 | (L<C<symlink>|perlfunc/symlink OLDFILE,NEWFILE>, | |
275 | L<C<readlink>|perlfunc/readlink EXPR>, | |
276 | L<C<lstat>|perlfunc/lstat FILEHANDLE>). | |
433acd8a | 277 | |
6ab3f9cb | 278 | The filesystem may support neither access timestamp nor change |
433acd8a JH |
279 | timestamp (meaning that about the only portable timestamp is the |
280 | modification timestamp), or one second granularity of any timestamps | |
281 | (e.g. the FAT filesystem limits the time granularity to two seconds). | |
282 | ||
83a46a63 LM |
283 | The "inode change timestamp" (the L<C<-C>|perlfunc/-X FILEHANDLE> |
284 | filetest) may really be the "creation timestamp" (which it is not in | |
285 | Unix). | |
95a3fe12 | 286 | |
495c5fdc PG |
287 | VOS perl can emulate Unix filenames with C</> as path separator. The |
288 | native pathname characters greater-than, less-than, number-sign, and | |
289 | percent-sign are always accepted. | |
290 | ||
6ab3f9cb | 291 | S<RISC OS> perl can emulate Unix filenames with C</> as path |
322422de | 292 | separator, or go native and use C<.> for path separator and C<:> to |
6ab3f9cb | 293 | signal filesystems and disk names. |
e41182b5 | 294 | |
e1020413 | 295 | Don't assume Unix filesystem access semantics: that read, write, |
a1667ba3 | 296 | and execute are all the permissions there are, and even if they exist, |
83a46a63 | 297 | that their semantics (for example what do C<r>, C<w>, and C<x> mean on |
e1020413 | 298 | a directory) are the Unix ones. The various Unix/POSIX compatibility |
83a46a63 LM |
299 | layers usually try to make interfaces like L<C<chmod>|perlfunc/chmod LIST> |
300 | work, but sometimes there simply is no good mapping. | |
a1667ba3 | 301 | |
83a46a63 | 302 | The L<C<File::Spec>|File::Spec> modules provide methods to manipulate path |
35a328a7 CB |
303 | specifications and return the results in native format for each |
304 | platform. This is often unnecessary as Unix-style paths are | |
305 | understood by Perl on every supported platform, but if you need to | |
306 | produce native paths for a native utility that does not understand | |
307 | Unix syntax, or if you are operating on paths or path components | |
83a46a63 LM |
308 | in unknown (and thus possibly native) syntax, L<C<File::Spec>|File::Spec> |
309 | is your friend. Here are two brief examples: | |
e41182b5 | 310 | |
6ab3f9cb GS |
311 | use File::Spec::Functions; |
312 | chdir(updir()); # go up one directory | |
35a328a7 CB |
313 | |
314 | # Concatenate a path from its components | |
315 | my $file = catfile(updir(), 'temp', 'file.txt'); | |
316 | # on Unix: '../temp/file.txt' | |
317 | # on Win32: '..\temp\file.txt' | |
318 | # on VMS: '[-.temp]file.txt' | |
e41182b5 | 319 | |
b7df3edc GS |
320 | In general, production code should not have file paths hardcoded. |
321 | Making them user-supplied or read from a configuration file is | |
322 | better, keeping in mind that file path syntax varies on different | |
323 | machines. | |
e41182b5 GS |
324 | |
325 | This is especially noticeable in scripts like Makefiles and test suites, | |
326 | which often assume C</> as a path separator for subdirectories. | |
327 | ||
83a46a63 LM |
328 | Also of use is L<C<File::Basename>|File::Basename> from the standard |
329 | distribution, which splits a pathname into pieces (base filename, full | |
330 | path to directory, and file suffix). | |
e41182b5 | 331 | |
19799a22 | 332 | Even when on a single platform (if you can call Unix a single platform), |
b7df3edc | 333 | remember not to count on the existence or the contents of particular |
3c075c7d | 334 | system-specific files or directories, like F</etc/passwd>, |
b7df3edc GS |
335 | F</etc/sendmail.conf>, F</etc/resolv.conf>, or even F</tmp/>. For |
336 | example, F</etc/passwd> may exist but not contain the encrypted | |
337 | passwords, because the system is using some form of enhanced security. | |
2c044526 | 338 | Or it may not contain all the accounts, because the system is using NIS. |
3c075c7d | 339 | If code does need to rely on such a file, include a description of the |
b7df3edc | 340 | file and its format in the code's documentation, then make it easy for |
3c075c7d CN |
341 | the user to override the default location of the file. |
342 | ||
b7df3edc GS |
343 | Don't assume a text file will end with a newline. They should, |
344 | but people forget. | |
e41182b5 | 345 | |
ec481373 JH |
346 | Do not have two files or directories of the same name with different |
347 | case, like F<test.pl> and F<Test.pl>, as many platforms have | |
348 | case-insensitive (or at least case-forgiving) filenames. Also, try | |
349 | not to have non-word characters (except for C<.>) in the names, and | |
350 | keep them to the 8.3 convention, for maximum portability, onerous a | |
351 | burden though this may appear. | |
dd9f0070 | 352 | |
83a46a63 LM |
353 | Likewise, when using the L<C<AutoSplit>|AutoSplit> module, try to keep |
354 | your functions to 8.3 naming and case-insensitive conventions; or, at the | |
355 | least, make it so the resulting files have a unique (case-insensitively) | |
dd9f0070 CN |
356 | first 8 characters. |
357 | ||
ec481373 JH |
358 | Whitespace in filenames is tolerated on most systems, but not all, |
359 | and even on systems where it might be tolerated, some utilities | |
fe829689 | 360 | might become confused by such whitespace. |
ec481373 | 361 | |
016930a6 JM |
362 | Many systems (DOS, VMS ODS-2) cannot have more than one C<.> in their |
363 | filenames. | |
433acd8a | 364 | |
c47ff5f1 | 365 | Don't assume C<< > >> won't be the first character of a filename. |
83a46a63 LM |
366 | Always use the three-arg version of |
367 | L<C<open>|perlfunc/open FILEHANDLE,EXPR>: | |
0a47030a | 368 | |
ceaffd1d | 369 | open my $fh, '<', $existing_file) or die $!; |
0a47030a | 370 | |
83a46a63 LM |
371 | Two-arg L<C<open>|perlfunc/open FILEHANDLE,EXPR> is magic and can |
372 | translate characters like C<< > >>, C<< < >>, and C<|> in filenames, | |
373 | which is usually the wrong thing to do. | |
374 | L<C<sysopen>|perlfunc/sysopen FILEHANDLE,FILENAME,MODE> and three-arg | |
375 | L<C<open>|perlfunc/open FILEHANDLE,EXPR> don't have this problem. | |
e41182b5 | 376 | |
ec481373 | 377 | Don't use C<:> as a part of a filename since many systems use that for |
8939ba94 | 378 | their own semantics (Mac OS Classic for separating pathname components, |
ec481373 | 379 | many networking schemes and utilities for separating the nodename and |
08fef530 JH |
380 | the pathname, and so on). For the same reasons, avoid C<@>, C<;> and |
381 | C<|>. | |
ec481373 | 382 | |
e1516da7 JH |
383 | Don't assume that in pathnames you can collapse two leading slashes |
384 | C<//> into one: some networking and clustering filesystems have special | |
2c044526 | 385 | semantics for that. Let the operating system sort it out. |
e1516da7 | 386 | |
ec481373 JH |
387 | The I<portable filename characters> as defined by ANSI C are |
388 | ||
1802421e JK |
389 | a b c d e f g h i j k l m n o p q r s t u v w x y z |
390 | A B C D E F G H I J K L M N O P Q R S T U V W X Y Z | |
ec481373 JH |
391 | 0 1 2 3 4 5 6 7 8 9 |
392 | . _ - | |
393 | ||
83a46a63 | 394 | and C<-> shouldn't be the first character. If you want to be |
e1516da7 JH |
395 | hypercorrect, stay case-insensitive and within the 8.3 naming |
396 | convention (all the files and directories have to be unique within one | |
397 | directory if their names are lowercased and truncated to eight | |
398 | characters before the C<.>, if any, and to three characters after the | |
399 | C<.>, if any). (And do not use C<.>s in directory names.) | |
ec481373 | 400 | |
e41182b5 GS |
401 | =head2 System Interaction |
402 | ||
b7df3edc GS |
403 | Not all platforms provide a command line. These are usually platforms |
404 | that rely primarily on a Graphical User Interface (GUI) for user | |
405 | interaction. A program requiring a command line interface might | |
406 | not work everywhere. This is probably for the user of the program | |
407 | to deal with, so don't stay up late worrying about it. | |
e41182b5 | 408 | |
c87488a3 HS |
409 | Some platforms can't delete or rename files held open by the system, |
410 | this limitation may also apply to changing filesystem metainformation | |
83a46a63 LM |
411 | like file permissions or owners. Remember to |
412 | L<C<close>|perlfunc/close FILEHANDLE> files when you are done with them. | |
413 | Don't L<C<unlink>|perlfunc/unlink LIST> or | |
414 | L<C<rename>|perlfunc/rename OLDNAME,NEWNAME> an open file. Don't | |
415 | L<C<tie>|perlfunc/tie VARIABLE,CLASSNAME,LIST> or | |
416 | L<C<open>|perlfunc/open FILEHANDLE,EXPR> a file already tied or opened; | |
417 | L<C<untie>|perlfunc/untie VARIABLE> or | |
418 | L<C<close>|perlfunc/close FILEHANDLE> it first. | |
e41182b5 | 419 | |
0a47030a GS |
420 | Don't open the same file more than once at a time for writing, as some |
421 | operating systems put mandatory locks on such files. | |
422 | ||
73e9292c JH |
423 | Don't assume that write/modify permission on a directory gives the |
424 | right to add or delete files/directories in that directory. That is | |
425 | filesystem specific: in some filesystems you need write/modify | |
426 | permission also (or even just) in the file/directory itself. In some | |
427 | filesystems (AFS, DFS) the permission to add/delete directory entries | |
428 | is a completely separate permission. | |
429 | ||
83a46a63 LM |
430 | Don't assume that a single L<C<unlink>|perlfunc/unlink LIST> completely |
431 | gets rid of the file: some filesystems (most notably the ones in VMS) have | |
432 | versioned filesystems, and L<C<unlink>|perlfunc/unlink LIST> removes only | |
433 | the most recent one (it doesn't remove all the versions because by default | |
434 | the native tools on those platforms remove just the most recent version, | |
435 | too). The portable idiom to remove all the versions of a file is | |
73e9292c | 436 | |
94bb614c | 437 | 1 while unlink "file"; |
73e9292c | 438 | |
dabde021 | 439 | This will terminate if the file is undeletable for some reason |
73e9292c JH |
440 | (protected, not there, and so on). |
441 | ||
83a46a63 LM |
442 | Don't count on a specific environment variable existing in |
443 | L<C<%ENV>|perlvar/%ENV>. Don't count on L<C<%ENV>|perlvar/%ENV> entries | |
444 | being case-sensitive, or even case-preserving. Don't try to clear | |
445 | L<C<%ENV>|perlvar/%ENV> by saying C<%ENV = ();>, or, if you really have | |
446 | to, make it conditional on C<$^O ne 'VMS'> since in VMS the | |
447 | L<C<%ENV>|perlvar/%ENV> table is much more than a per-process key-value | |
448 | string table. | |
449 | ||
450 | On VMS, some entries in the L<C<%ENV>|perlvar/%ENV> hash are dynamically | |
451 | created when their key is used on a read if they did not previously | |
452 | exist. The values for C<$ENV{HOME}>, C<$ENV{TERM}>, C<$ENV{PATH}>, and | |
453 | C<$ENV{USER}>, are known to be dynamically generated. The specific names | |
454 | that are dynamically generated may vary with the version of the C library | |
455 | on VMS, and more may exist than are documented. | |
456 | ||
457 | On VMS by default, changes to the L<C<%ENV>|perlvar/%ENV> hash persist | |
458 | after perl exits. Subsequent invocations of perl in the same process can | |
459 | inadvertently inherit environment settings that were meant to be | |
460 | temporary. | |
461 | ||
462 | Don't count on signals or L<C<%SIG>|perlvar/%SIG> for anything. | |
463 | ||
464 | Don't count on filename globbing. Use | |
465 | L<C<opendir>|perlfunc/opendir DIRHANDLE,EXPR>, | |
466 | L<C<readdir>|perlfunc/readdir DIRHANDLE>, and | |
467 | L<C<closedir>|perlfunc/closedir DIRHANDLE> instead. | |
e41182b5 | 468 | |
b8099c3d | 469 | Don't count on per-program environment variables, or per-program current |
dd9f0070 | 470 | directories. |
b8099c3d | 471 | |
83a46a63 | 472 | Don't count on specific values of L<C<$!>|perlvar/$!>, neither numeric nor |
2c044526 | 473 | especially the string values. Users may switch their locales causing |
c87488a3 HS |
474 | error messages to be translated into their languages. If you can |
475 | trust a POSIXish environment, you can portably use the symbols defined | |
83a46a63 LM |
476 | by the L<C<Errno>|Errno> module, like C<ENOENT>. And don't trust on the |
477 | values of L<C<$!>|perlvar/$!> at all except immediately after a failed | |
478 | system call. | |
3c075c7d | 479 | |
a10d74f3 PG |
480 | =head2 Command names versus file pathnames |
481 | ||
482 | Don't assume that the name used to invoke a command or program with | |
83a46a63 LM |
483 | L<C<system>|perlfunc/system LIST> or L<C<exec>|perlfunc/exec LIST> can |
484 | also be used to test for the existence of the file that holds the | |
485 | executable code for that command or program. | |
68fb0eb7 PG |
486 | First, many systems have "internal" commands that are built-in to the |
487 | shell or OS and while these commands can be invoked, there is no | |
488 | corresponding file. Second, some operating systems (e.g., Cygwin, | |
489 | DJGPP, OS/2, and VOS) have required suffixes for executable files; | |
490 | these suffixes are generally permitted on the command name but are not | |
83a46a63 LM |
491 | required. Thus, a command like C<perl> might exist in a file named |
492 | F<perl>, F<perl.exe>, or F<perl.pm>, depending on the operating system. | |
493 | The variable L<C<$Config{_exe}>|Config/C<_exe>> in the | |
494 | L<C<Config>|Config> module holds the executable suffix, if any. Third, | |
495 | the VMS port carefully sets up L<C<$^X>|perlvar/$^X> and | |
496 | L<C<$Config{perlpath}>|Config/C<perlpath>> so that no further processing | |
497 | is required. This is just as well, because the matching regular | |
498 | expression used below would then have to deal with a possible trailing | |
499 | version number in the VMS file name. | |
500 | ||
501 | To convert L<C<$^X>|perlvar/$^X> to a file pathname, taking account of | |
502 | the requirements of the various operating system possibilities, say: | |
7ee27b7c | 503 | |
a61fc69c KW |
504 | use Config; |
505 | my $thisperl = $^X; | |
83a46a63 LM |
506 | if ($^O ne 'VMS') { |
507 | $thisperl .= $Config{_exe} | |
508 | unless $thisperl =~ m/\Q$Config{_exe}\E$/i; | |
509 | } | |
a10d74f3 | 510 | |
83a46a63 | 511 | To convert L<C<$Config{perlpath}>|Config/C<perlpath>> to a file pathname, say: |
7ee27b7c | 512 | |
a61fc69c KW |
513 | use Config; |
514 | my $thisperl = $Config{perlpath}; | |
83a46a63 LM |
515 | if ($^O ne 'VMS') { |
516 | $thisperl .= $Config{_exe} | |
517 | unless $thisperl =~ m/\Q$Config{_exe}\E$/i; | |
518 | } | |
a10d74f3 | 519 | |
7137b697 JH |
520 | =head2 Networking |
521 | ||
522 | Don't assume that you can reach the public Internet. | |
523 | ||
524 | Don't assume that there is only one way to get through firewalls | |
525 | to the public Internet. | |
526 | ||
932f293e JH |
527 | Don't assume that you can reach outside world through any other port |
528 | than 80, or some web proxy. ftp is blocked by many firewalls. | |
529 | ||
dbc6a9ce JH |
530 | Don't assume that you can send email by connecting to the local SMTP port. |
531 | ||
7137b697 | 532 | Don't assume that you can reach yourself or any node by the name |
dbc6a9ce | 533 | 'localhost'. The same goes for '127.0.0.1'. You will have to try both. |
932f293e | 534 | |
86feb2c5 JH |
535 | Don't assume that the host has only one network card, or that it |
536 | can't bind to many virtual IP addresses. | |
932f293e JH |
537 | |
538 | Don't assume a particular network device name. | |
7137b697 | 539 | |
83a46a63 LM |
540 | Don't assume a particular set of |
541 | L<C<ioctl>|perlfunc/ioctl FILEHANDLE,FUNCTION,SCALAR>s will work. | |
7137b697 JH |
542 | |
543 | Don't assume that you can ping hosts and get replies. | |
544 | ||
dbc6a9ce JH |
545 | Don't assume that any particular port (service) will respond. |
546 | ||
83a46a63 LM |
547 | Don't assume that L<C<Sys::Hostname>|Sys::Hostname> (or any other API or |
548 | command) returns either a fully qualified hostname or a non-qualified | |
549 | hostname: it all depends on how the system had been configured. Also | |
550 | remember that for things such as DHCP and NAT, the hostname you get back | |
551 | might not be very useful. | |
dbc6a9ce | 552 | |
a746ef5c | 553 | All the above I<don't>s may look daunting, and they are, but the key |
932f293e JH |
554 | is to degrade gracefully if one cannot reach the particular network |
555 | service one wants. Croaking or hanging do not look very professional. | |
556 | ||
e41182b5 GS |
557 | =head2 Interprocess Communication (IPC) |
558 | ||
b7df3edc | 559 | In general, don't directly access the system in code meant to be |
83a46a63 LM |
560 | portable. That means, no L<C<system>|perlfunc/system LIST>, |
561 | L<C<exec>|perlfunc/exec LIST>, L<C<fork>|perlfunc/fork>, | |
562 | L<C<pipe>|perlfunc/pipe READHANDLE,WRITEHANDLE>, | |
563 | L<C<``> or C<qxE<sol>E<sol>>|perlop/C<qxE<sol>I<STRING>E<sol>>>, | |
564 | L<C<open>|perlfunc/open FILEHANDLE,EXPR> with a C<|>, nor any of the other | |
565 | things that makes being a Perl hacker worth being. | |
e41182b5 GS |
566 | |
567 | Commands that launch external processes are generally supported on | |
b7df3edc GS |
568 | most platforms (though many of them do not support any type of |
569 | forking). The problem with using them arises from what you invoke | |
570 | them on. External tools are often named differently on different | |
4375e838 | 571 | platforms, may not be available in the same location, might accept |
b7df3edc GS |
572 | different arguments, can behave differently, and often present their |
573 | results in a platform-dependent way. Thus, you should seldom depend | |
83a46a63 LM |
574 | on them to produce consistent results. (Then again, if you're calling |
575 | C<netstat -a>, you probably don't expect it to run on both Unix and CP/M.) | |
e41182b5 | 576 | |
b7df3edc | 577 | One especially common bit of Perl code is opening a pipe to B<sendmail>: |
e41182b5 | 578 | |
83a46a63 | 579 | open(my $mail, '|-', '/usr/lib/sendmail -t') |
b7df3edc | 580 | or die "cannot fork sendmail: $!"; |
e41182b5 GS |
581 | |
582 | This is fine for systems programming when sendmail is known to be | |
583 | available. But it is not fine for many non-Unix systems, and even | |
584 | some Unix systems that may not have sendmail installed. If a portable | |
b7df3edc | 585 | solution is needed, see the various distributions on CPAN that deal |
83a46a63 LM |
586 | with it. L<C<Mail::Mailer>|Mail::Mailer> and L<C<Mail::Send>|Mail::Send> |
587 | in the C<MailTools> distribution are commonly used, and provide several | |
588 | mailing methods, including C<mail>, C<sendmail>, and direct SMTP (via | |
589 | L<C<Net::SMTP>|Net::SMTP>) if a mail transfer agent is not available. | |
590 | L<C<Mail::Sendmail>|Mail::Sendmail> is a standalone module that provides | |
b7df3edc GS |
591 | simple, platform-independent mailing. |
592 | ||
593 | The Unix System V IPC (C<msg*(), sem*(), shm*()>) is not available | |
594 | even on all Unix platforms. | |
e41182b5 | 595 | |
a81e5e2e A |
596 | Do not use either the bare result of C<pack("N", 10, 20, 30, 40)> or |
597 | bare v-strings (such as C<v10.20.30.40>) to represent IPv4 addresses: | |
598 | both forms just pack the four bytes into network order. That this | |
599 | would be equal to the C language C<in_addr> struct (which is what the | |
600 | socket code internally uses) is not guaranteed. To be portable use | |
83a46a63 LM |
601 | the routines of the L<C<Socket>|Socket> module, such as |
602 | L<C<inet_aton>|Socket/$ip_address = inet_aton $string>, | |
603 | L<C<inet_ntoa>|Socket/$string = inet_ntoa $ip_address>, and | |
604 | L<C<sockaddr_in>|Socket/$sockaddr = sockaddr_in $port, $ip_address>. | |
6b2463a0 | 605 | |
e41182b5 | 606 | The rule of thumb for portable code is: Do it all in portable Perl, or |
0a47030a | 607 | use a module (that may internally implement it with platform-specific |
2c044526 | 608 | code, but exposes a common interface). |
e41182b5 | 609 | |
e41182b5 GS |
610 | =head2 External Subroutines (XS) |
611 | ||
b7df3edc | 612 | XS code can usually be made to work with any platform, but dependent |
e41182b5 GS |
613 | libraries, header files, etc., might not be readily available or |
614 | portable, or the XS code itself might be platform-specific, just as Perl | |
615 | code might be. If the libraries and headers are portable, then it is | |
616 | normally reasonable to make sure the XS code is portable, too. | |
617 | ||
b7df3edc GS |
618 | A different type of portability issue arises when writing XS code: |
619 | availability of a C compiler on the end-user's system. C brings | |
620 | with it its own portability issues, and writing XS code will expose | |
621 | you to some of those. Writing purely in Perl is an easier way to | |
e41182b5 GS |
622 | achieve portability. |
623 | ||
e41182b5 GS |
624 | =head2 Standard Modules |
625 | ||
626 | In general, the standard modules work across platforms. Notable | |
83a46a63 LM |
627 | exceptions are the L<C<CPAN>|CPAN> module (which currently makes |
628 | connections to external programs that may not be available), | |
629 | platform-specific modules (like L<C<ExtUtils::MM_VMS>|ExtUtils::MM_VMS>), | |
630 | and DBM modules. | |
e41182b5 | 631 | |
b7df3edc | 632 | There is no one DBM module available on all platforms. |
83a46a63 LM |
633 | L<C<SDBM_File>|SDBM_File> and the others are generally available on all |
634 | Unix and DOSish ports, but not in MacPerl, where only | |
635 | L<C<NDBM_File>|NDBM_File> and L<C<DB_File>|DB_File> are available. | |
e41182b5 GS |
636 | |
637 | The good news is that at least some DBM module should be available, and | |
83a46a63 LM |
638 | L<C<AnyDBM_File>|AnyDBM_File> will use whichever module it can find. Of |
639 | course, then the code needs to be fairly strict, dropping to the greatest | |
640 | common factor (e.g., not exceeding 1K for each record), so that it will | |
6ab3f9cb | 641 | work with any DBM module. See L<AnyDBM_File> for more details. |
e41182b5 | 642 | |
e41182b5 GS |
643 | =head2 Time and Date |
644 | ||
0a47030a | 645 | The system's notion of time of day and calendar date is controlled in |
b7df3edc | 646 | widely different ways. Don't assume the timezone is stored in C<$ENV{TZ}>, |
0a47030a | 647 | and even if it is, don't assume that you can control the timezone through |
c87488a3 HS |
648 | that variable. Don't assume anything about the three-letter timezone |
649 | abbreviations (for example that MST would be the Mountain Standard Time, | |
650 | it's been known to stand for Moscow Standard Time). If you need to | |
651 | use timezones, express them in some unambiguous format like the | |
652 | exact number of minutes offset from UTC, or the POSIX timezone | |
653 | format. | |
e41182b5 | 654 | |
322422de | 655 | Don't assume that the epoch starts at 00:00:00, January 1, 1970, |
c87488a3 HS |
656 | because that is OS- and implementation-specific. It is better to |
657 | store a date in an unambiguous representation. The ISO 8601 standard | |
766af94f | 658 | defines YYYY-MM-DD as the date format, or YYYY-MM-DDTHH:MM:SS |
c87488a3 | 659 | (that's a literal "T" separating the date from the time). |
91d20606 | 660 | Please do use the ISO 8601 instead of making us guess what |
c87488a3 HS |
661 | date 02/03/04 might be. ISO 8601 even sorts nicely as-is. |
662 | A text representation (like "1987-12-18") can be easily converted | |
83a46a63 LM |
663 | into an OS-specific value using a module like |
664 | L<C<Time::Piece>|Time::Piece> (see L<Time::Piece/Date Parsing>) or | |
665 | L<C<Date::Parse>|Date::Parse>. An array of values, such as those | |
666 | returned by L<C<localtime>|perlfunc/localtime EXPR>, can be converted to an OS-specific | |
667 | representation using L<C<Time::Local>|Time::Local>. | |
322422de | 668 | |
19799a22 GS |
669 | When calculating specific times, such as for tests in time or date modules, |
670 | it may be appropriate to calculate an offset for the epoch. | |
b7df3edc | 671 | |
83a46a63 LM |
672 | use Time::Local qw(timegm); |
673 | my $offset = timegm(0, 0, 0, 1, 0, 70); | |
b7df3edc | 674 | |
204ad8d5 JV |
675 | The value for C<$offset> in Unix will be C<0>, but in Mac OS Classic |
676 | will be some large number. C<$offset> can then be added to a Unix time | |
677 | value to get what should be the proper value on any system. | |
322422de GS |
678 | |
679 | =head2 Character sets and character encoding | |
680 | ||
ec481373 JH |
681 | Assume very little about character sets. |
682 | ||
83a46a63 LM |
683 | Assume nothing about numerical values (L<C<ord>|perlfunc/ord EXPR>, |
684 | L<C<chr>|perlfunc/chr NUMBER>) of characters. | |
eb9df707 KW |
685 | Do not use explicit code point ranges (like C<\xHH-\xHH)>. However, |
686 | starting in Perl v5.22, regular expression pattern bracketed character | |
f4240379 | 687 | class ranges specified like C<qr/[\N{U+HH}-\N{U+HH}]/> are portable, |
83a46a63 LM |
688 | and starting in Perl v5.24, the same ranges are portable in |
689 | L<C<trE<sol>E<sol>E<sol>>|perlop/C<trE<sol>I<SEARCHLIST>E<sol>I<REPLACEMENTLIST>E<sol>cdsr>>. | |
2c044526 | 690 | You can portably use symbolic character classes like C<[:print:]>. |
ec481373 JH |
691 | |
692 | Do not assume that the alphabetic characters are encoded contiguously | |
eb9df707 KW |
693 | (in the numeric sense). There may be gaps. Special coding in Perl, |
694 | however, guarantees that all subsets of C<qr/[A-Z]/>, C<qr/[a-z]/>, and | |
83a46a63 LM |
695 | C<qr/[0-9]/> behave as expected. |
696 | L<C<trE<sol>E<sol>E<sol>>|perlop/C<trE<sol>I<SEARCHLIST>E<sol>I<REPLACEMENTLIST>E<sol>cdsr>> | |
697 | behaves the same for these ranges. In patterns, any ranges specified with | |
698 | end points using the C<\N{...}> notations ensures character set | |
699 | portability, but it is a bug in Perl v5.22 that this isn't true of | |
700 | L<C<trE<sol>E<sol>E<sol>>|perlop/C<trE<sol>I<SEARCHLIST>E<sol>I<REPLACEMENTLIST>E<sol>cdsr>>, | |
701 | fixed in v5.24. | |
ec481373 JH |
702 | |
703 | Do not assume anything about the ordering of the characters. | |
704 | The lowercase letters may come before or after the uppercase letters; | |
b432a672 AL |
705 | the lowercase and uppercase may be interlaced so that both "a" and "A" |
706 | come before "b"; the accented and other international characters may | |
707 | be interlaced so that E<auml> comes before "b". | |
2c044526 | 708 | L<Unicode::Collate> can be used to sort this all out. |
322422de GS |
709 | |
710 | =head2 Internationalisation | |
711 | ||
b7df3edc GS |
712 | If you may assume POSIX (a rather large assumption), you may read |
713 | more about the POSIX locale system from L<perllocale>. The locale | |
714 | system at least attempts to make things a little bit more portable, | |
715 | or at least more convenient and native-friendly for non-English | |
716 | users. The system affects character sets and encoding, and date | |
717 | and time formatting--amongst other things. | |
e41182b5 | 718 | |
c87488a3 HS |
719 | If you really want to be international, you should consider Unicode. |
720 | See L<perluniintro> and L<perlunicode> for more information. | |
721 | ||
83a46a63 LM |
722 | By default Perl assumes your source code is written in an 8-bit ASCII |
723 | superset. To embed Unicode characters in your strings and regexes, you can | |
724 | use the L<C<\x{HH}> or (more portably) C<\N{U+HH}> | |
725 | notations|perlop/Quote and Quote-like Operators>. You can also use the | |
726 | L<C<utf8>|utf8> pragma and write your code in UTF-8, which lets you use | |
727 | Unicode characters directly (not just in quoted constructs but also in | |
728 | identifiers). | |
11264fdb | 729 | |
e41182b5 GS |
730 | =head2 System Resources |
731 | ||
0a47030a GS |
732 | If your code is destined for systems with severely constrained (or |
733 | missing!) virtual memory systems then you want to be I<especially> mindful | |
734 | of avoiding wasteful constructs such as: | |
e41182b5 | 735 | |
ceaffd1d | 736 | my @lines = <$very_large_file>; # bad |
e41182b5 | 737 | |
ceaffd1d | 738 | while (<$fh>) {$file .= $_} # sometimes bad |
739 | my $file = join('', <$fh>); # better | |
e41182b5 | 740 | |
b7df3edc GS |
741 | The last two constructs may appear unintuitive to most people. The |
742 | first repeatedly grows a string, whereas the second allocates a | |
743 | large chunk of memory in one go. On some systems, the second is | |
2c044526 | 744 | more efficient than the first. |
0a47030a | 745 | |
e41182b5 GS |
746 | =head2 Security |
747 | ||
b7df3edc | 748 | Most multi-user platforms provide basic levels of security, usually |
ac036724 | 749 | implemented at the filesystem level. Some, however, unfortunately do |
750 | not. Thus the notion of user id, or "home" directory, | |
b7df3edc GS |
751 | or even the state of being logged-in, may be unrecognizable on many |
752 | platforms. If you write programs that are security-conscious, it | |
753 | is usually best to know what type of system you will be running | |
754 | under so that you can write code explicitly for that platform (or | |
755 | class of platforms). | |
0a47030a | 756 | |
e1020413 | 757 | Don't assume the Unix filesystem access semantics: the operating |
a1667ba3 | 758 | system or the filesystem may be using some ACL systems, which are |
2c044526 | 759 | richer languages than the usual C<rwx>. Even if the C<rwx> exist, |
a1667ba3 JH |
760 | their semantics might be different. |
761 | ||
2c044526 | 762 | (From the security viewpoint, testing for permissions before attempting to |
a1667ba3 | 763 | do something is silly anyway: if one tries this, there is potential |
ac036724 | 764 | for race conditions. Someone or something might change the |
a1667ba3 JH |
765 | permissions between the permissions check and the actual operation. |
766 | Just try the operation.) | |
767 | ||
e1020413 | 768 | Don't assume the Unix user and group semantics: especially, don't |
83a46a63 LM |
769 | expect L<C<< $< >>|perlvar/$E<lt>> and L<C<< $> >>|perlvar/$E<gt>> (or |
770 | L<C<$(>|perlvar/$(> and L<C<$)>|perlvar/$)>) to work for switching | |
771 | identities (or memberships). | |
a1667ba3 | 772 | |
83a46a63 | 773 | Don't assume set-uid and set-gid semantics. (And even if you do, |
a1667ba3 JH |
774 | think twice: set-uid and set-gid are a known can of security worms.) |
775 | ||
e41182b5 GS |
776 | =head2 Style |
777 | ||
778 | For those times when it is necessary to have platform-specific code, | |
779 | consider keeping the platform-specific code in one place, making porting | |
83a46a63 LM |
780 | to other platforms easier. Use the L<C<Config>|Config> module and the |
781 | special variable L<C<$^O>|perlvar/$^O> to differentiate platforms, as | |
782 | described in L</"PLATFORMS">. | |
e41182b5 | 783 | |
d4c800c7 JH |
784 | Beware of the "else syndrome": |
785 | ||
786 | if ($^O eq 'MSWin32') { | |
787 | # code that assumes Windows | |
788 | } else { | |
789 | # code that assumes Linux | |
790 | } | |
791 | ||
792 | The C<else> branch should be used for the really ultimate fallback, | |
793 | not for code specific to some platform. | |
794 | ||
b7df3edc GS |
795 | Be careful in the tests you supply with your module or programs. |
796 | Module code may be fully portable, but its tests might not be. This | |
797 | often happens when tests spawn off other processes or call external | |
798 | programs to aid in the testing, or when (as noted above) the tests | |
c87488a3 HS |
799 | assume certain things about the filesystem and paths. Be careful not |
800 | to depend on a specific output style for errors, such as when checking | |
83a46a63 LM |
801 | L<C<$!>|perlvar/$!> after a failed system call. Using |
802 | L<C<$!>|perlvar/$!> for anything else than displaying it as output is | |
803 | doubtful (though see the L<C<Errno>|Errno> module for testing reasonably | |
804 | portably for error value). Some platforms expect a certain output format, | |
805 | and Perl on those platforms may have been adjusted accordingly. Most | |
806 | specifically, don't anchor a regex when testing an error value. | |
e41182b5 | 807 | |
0a47030a | 808 | =head1 CPAN Testers |
e41182b5 | 809 | |
0a47030a GS |
810 | Modules uploaded to CPAN are tested by a variety of volunteers on |
811 | different platforms. These CPAN testers are notified by mail of each | |
e41182b5 | 812 | new upload, and reply to the list with PASS, FAIL, NA (not applicable to |
0a47030a | 813 | this platform), or UNKNOWN (unknown), along with any relevant notations. |
e41182b5 GS |
814 | |
815 | The purpose of the testing is twofold: one, to help developers fix any | |
0a47030a | 816 | problems in their code that crop up because of lack of testing on other |
b7df3edc | 817 | platforms; two, to provide users with information about whether |
0a47030a | 818 | a given module works on a given platform. |
e41182b5 | 819 | |
2890cc8c | 820 | Also see: |
7ee27b7c | 821 | |
e41182b5 GS |
822 | =over 4 |
823 | ||
7ee27b7c AT |
824 | =item * |
825 | ||
636280bd | 826 | Mailing list: cpan-testers-discuss@perl.org |
7ee27b7c AT |
827 | |
828 | =item * | |
e41182b5 | 829 | |
500f1b69 | 830 | Testing results: L<http://www.cpantesters.org/> |
e41182b5 GS |
831 | |
832 | =back | |
833 | ||
e41182b5 GS |
834 | =head1 PLATFORMS |
835 | ||
83a46a63 LM |
836 | Perl is built with a L<C<$^O>|perlvar/$^O> variable that indicates the |
837 | operating system it was built on. This was implemented | |
b7df3edc | 838 | to help speed up code that would otherwise have to C<use Config> |
83a46a63 LM |
839 | and use the value of L<C<$Config{osname}>|Config/C<osname>>. Of course, |
840 | to get more detailed information about the system, looking into | |
841 | L<C<%Config>|Config/DESCRIPTION> is certainly recommended. | |
e41182b5 | 842 | |
83a46a63 LM |
843 | L<C<%Config>|Config/DESCRIPTION> cannot always be trusted, however, |
844 | because it was built at compile time. If perl was built in one place, | |
845 | then transferred elsewhere, some values may be wrong. The values may | |
846 | even have been edited after the fact. | |
6ab3f9cb | 847 | |
e41182b5 GS |
848 | =head2 Unix |
849 | ||
850 | Perl works on a bewildering variety of Unix and Unix-like platforms (see | |
851 | e.g. most of the files in the F<hints/> directory in the source code kit). | |
83a46a63 LM |
852 | On most of these systems, the value of L<C<$^O>|perlvar/$^O> (hence |
853 | L<C<$Config{osname}>|Config/C<osname>>, too) is determined either by | |
854 | lowercasing and stripping punctuation from the first field of the string | |
855 | returned by typing C<uname -a> (or a similar command) at the shell prompt | |
856 | or by testing the file system for the presence of uniquely named files | |
857 | such as a kernel or header file. Here, for example, are a few of the | |
858 | more popular Unix flavors: | |
859 | ||
860 | uname $^O $Config{archname} | |
6ab3f9cb | 861 | -------------------------------------------- |
b7df3edc | 862 | AIX aix aix |
6ab3f9cb | 863 | BSD/OS bsdos i386-bsdos |
e1516da7 | 864 | Darwin darwin darwin |
6ab3f9cb | 865 | DYNIX/ptx dynixptx i386-dynixptx |
2890cc8c | 866 | FreeBSD freebsd freebsd-i386 |
df00ff3b | 867 | Haiku haiku BePC-haiku |
d1e3b762 | 868 | Linux linux arm-linux |
56b575b9 | 869 | Linux linux armv5tel-linux |
b7df3edc | 870 | Linux linux i386-linux |
6ab3f9cb GS |
871 | Linux linux i586-linux |
872 | Linux linux ppc-linux | |
b7df3edc GS |
873 | HP-UX hpux PA-RISC1.1 |
874 | IRIX irix irix | |
b787fad4 | 875 | Mac OS X darwin darwin |
d1e3b762 GS |
876 | NeXT 3 next next-fat |
877 | NeXT 4 next OPENSTEP-Mach | |
6ab3f9cb | 878 | openbsd openbsd i386-openbsd |
b7df3edc | 879 | OSF1 dec_osf alpha-dec_osf |
6ab3f9cb GS |
880 | reliantunix-n svr4 RM400-svr4 |
881 | SCO_SV sco_sv i386-sco_sv | |
882 | SINIX-N svr4 RM400-svr4 | |
883 | sn4609 unicos CRAY_C90-unicos | |
884 | sn6521 unicosmk t3e-unicosmk | |
885 | sn9617 unicos CRAY_J90-unicos | |
b7df3edc GS |
886 | SunOS solaris sun4-solaris |
887 | SunOS solaris i86pc-solaris | |
888 | SunOS4 sunos sun4-sunos | |
e41182b5 | 889 | |
83a46a63 LM |
890 | Because the value of L<C<$Config{archname}>|Config/C<archname>> may |
891 | depend on the hardware architecture, it can vary more than the value of | |
892 | L<C<$^O>|perlvar/$^O>. | |
6ab3f9cb | 893 | |
e41182b5 GS |
894 | =head2 DOS and Derivatives |
895 | ||
b7df3edc | 896 | Perl has long been ported to Intel-style microcomputers running under |
e41182b5 GS |
897 | systems like PC-DOS, MS-DOS, OS/2, and most Windows platforms you can |
898 | bring yourself to mention (except for Windows CE, if you count that). | |
b7df3edc | 899 | Users familiar with I<COMMAND.COM> or I<CMD.EXE> style shells should |
e41182b5 GS |
900 | be aware that each of these file specifications may have subtle |
901 | differences: | |
902 | ||
ceaffd1d | 903 | my $filespec0 = "c:/foo/bar/file.txt"; |
904 | my $filespec1 = "c:\\foo\\bar\\file.txt"; | |
905 | my $filespec2 = 'c:\foo\bar\file.txt'; | |
906 | my $filespec3 = 'c:\\foo\\bar\\file.txt'; | |
e41182b5 | 907 | |
b7df3edc GS |
908 | System calls accept either C</> or C<\> as the path separator. |
909 | However, many command-line utilities of DOS vintage treat C</> as | |
910 | the option prefix, so may get confused by filenames containing C</>. | |
911 | Aside from calling any external programs, C</> will work just fine, | |
912 | and probably better, as it is more consistent with popular usage, | |
913 | and avoids the problem of remembering what to backwhack and what | |
914 | not to. | |
e41182b5 | 915 | |
b7df3edc GS |
916 | The DOS FAT filesystem can accommodate only "8.3" style filenames. Under |
917 | the "case-insensitive, but case-preserving" HPFS (OS/2) and NTFS (NT) | |
0a47030a | 918 | filesystems you may have to be careful about case returned with functions |
83a46a63 LM |
919 | like L<C<readdir>|perlfunc/readdir DIRHANDLE> or used with functions like |
920 | L<C<open>|perlfunc/open FILEHANDLE,EXPR> or | |
921 | L<C<opendir>|perlfunc/opendir DIRHANDLE,EXPR>. | |
e41182b5 | 922 | |
83a46a63 LM |
923 | DOS also treats several filenames as special, such as F<AUX>, F<PRN>, |
924 | F<NUL>, F<CON>, F<COM1>, F<LPT1>, F<LPT2>, etc. Unfortunately, sometimes | |
925 | these filenames won't even work if you include an explicit directory | |
926 | prefix. It is best to avoid such filenames, if you want your code to be | |
927 | portable to DOS and its derivatives. It's hard to know what these all | |
928 | are, unfortunately. | |
e41182b5 GS |
929 | |
930 | Users of these operating systems may also wish to make use of | |
83a46a63 LM |
931 | scripts such as F<pl2bat.bat> to put wrappers around your scripts. |
932 | ||
933 | Newline (C<\n>) is translated as C<\015\012> by the I/O system when | |
934 | reading from and writing to files (see L</"Newlines">). | |
935 | C<binmode($filehandle)> will keep C<\n> translated as C<\012> for that | |
936 | filehandle. | |
937 | L<C<binmode>|perlfunc/binmode FILEHANDLE> should always be used for code | |
938 | that deals with binary data. That's assuming you realize in advance that | |
939 | your data is in binary. General-purpose programs should often assume | |
940 | nothing about their data. | |
941 | ||
942 | The L<C<$^O>|perlvar/$^O> variable and the | |
943 | L<C<$Config{archname}>|Config/C<archname>> values for various DOSish | |
944 | perls are as follows: | |
945 | ||
946 | OS $^O $Config{archname} ID Version | |
947 | --------------------------------------------------------- | |
948 | MS-DOS dos ? | |
949 | PC-DOS dos ? | |
950 | OS/2 os2 ? | |
951 | Windows 3.1 ? ? 0 3 01 | |
952 | Windows 95 MSWin32 MSWin32-x86 1 4 00 | |
953 | Windows 98 MSWin32 MSWin32-x86 1 4 10 | |
954 | Windows ME MSWin32 MSWin32-x86 1 ? | |
955 | Windows NT MSWin32 MSWin32-x86 2 4 xx | |
956 | Windows NT MSWin32 MSWin32-ALPHA 2 4 xx | |
957 | Windows NT MSWin32 MSWin32-ppc 2 4 xx | |
958 | Windows 2000 MSWin32 MSWin32-x86 2 5 00 | |
959 | Windows XP MSWin32 MSWin32-x86 2 5 01 | |
960 | Windows 2003 MSWin32 MSWin32-x86 2 5 02 | |
961 | Windows Vista MSWin32 MSWin32-x86 2 6 00 | |
962 | Windows 7 MSWin32 MSWin32-x86 2 6 01 | |
963 | Windows 7 MSWin32 MSWin32-x64 2 6 01 | |
964 | Windows 2008 MSWin32 MSWin32-x86 2 6 01 | |
965 | Windows 2008 MSWin32 MSWin32-x64 2 6 01 | |
966 | Windows CE MSWin32 ? 3 | |
967 | Cygwin cygwin cygwin | |
e41182b5 | 968 | |
34aaaa84 | 969 | The various MSWin32 Perl's can distinguish the OS they are running on |
2890cc8c | 970 | via the value of the fifth element of the list returned from |
83a46a63 | 971 | L<C<Win32::GetOSVersion()>|Win32/Win32::GetOSVersion()>. For example: |
34aaaa84 PP |
972 | |
973 | if ($^O eq 'MSWin32') { | |
974 | my @os_version_info = Win32::GetOSVersion(); | |
975 | print +('3.1','95','NT')[$os_version_info[4]],"\n"; | |
976 | } | |
977 | ||
83a46a63 LM |
978 | There are also C<Win32::IsWinNT()|Win32/Win32::IsWinNT()>, |
979 | C<Win32::IsWin95()|Win32/Win32::IsWin95()>, and | |
980 | L<C<Win32::GetOSName()>|Win32/Win32::GetOSName()>; try | |
981 | L<C<perldoc Win32>|Win32>. | |
982 | The very portable L<C<POSIX::uname()>|POSIX/C<uname>> will work too: | |
1d65be3a JH |
983 | |
984 | c:\> perl -MPOSIX -we "print join '|', uname" | |
985 | Windows NT|moonru|5.0|Build 2195 (Service Pack 2)|x86 | |
d99f392e | 986 | |
1d061843 A |
987 | Errors set by Winsock functions are now put directly into C<$^E>, |
988 | and the relevant C<WSAE*> error codes are now exported from the | |
989 | L<Errno> and L<POSIX> modules for testing this against. | |
990 | ||
991 | The previous behavior of putting the errors (converted to POSIX-style | |
992 | C<E*> error codes since Perl 5.20.0) into C<$!> was buggy due to | |
993 | the non-equivalence of like-named Winsock and POSIX error constants, | |
994 | a relationship between which has unfortunately been established | |
995 | in one way or another since Perl 5.8.0. | |
996 | ||
997 | The new behavior provides a much more robust solution for checking | |
998 | Winsock errors in portable software without accidentally matching | |
999 | POSIX tests that were intended for other OSes and may have different | |
1000 | meanings for Winsock. | |
1001 | ||
1002 | The old behavior is currently retained, warts and all, for backwards | |
1003 | compatibility, but users are encouraged to change any code that | |
1004 | tests C<$!> against C<E*> constants for Winsock errors to instead | |
1005 | test C<$^E> against C<WSAE*> constants. After a suitable deprecation | |
1006 | period, which started with Perl 5.24, the old behavior may be | |
1007 | removed, leaving C<$!> unchanged after Winsock function calls, to | |
1008 | avoid any possible confusion over which error variable to check. | |
1009 | ||
e41182b5 GS |
1010 | Also see: |
1011 | ||
1012 | =over 4 | |
1013 | ||
c997b287 | 1014 | =item * |
e41182b5 | 1015 | |
500f1b69 | 1016 | The djgpp environment for DOS, L<http://www.delorie.com/djgpp/> |
c997b287 | 1017 | and L<perldos>. |
e41182b5 | 1018 | |
c997b287 | 1019 | =item * |
e41182b5 | 1020 | |
c997b287 | 1021 | The EMX environment for DOS, OS/2, etc. emx@iaehv.nl, |
500f1b69 | 1022 | L<ftp://hobbes.nmsu.edu/pub/os2/dev/emx/> Also L<perlos2>. |
e41182b5 | 1023 | |
c997b287 | 1024 | =item * |
d1e3b762 | 1025 | |
c997b287 | 1026 | Build instructions for Win32 in L<perlwin32>, or under the Cygnus environment |
2890cc8c | 1027 | in L<perlcygwin>. |
c997b287 GS |
1028 | |
1029 | =item * | |
1030 | ||
1031 | The C<Win32::*> modules in L<Win32>. | |
1032 | ||
1033 | =item * | |
1034 | ||
500f1b69 | 1035 | The ActiveState Pages, L<http://www.activestate.com/> |
c997b287 GS |
1036 | |
1037 | =item * | |
1038 | ||
2890cc8c | 1039 | The Cygwin environment for Win32; F<README.cygwin> (installed |
500f1b69 | 1040 | as L<perlcygwin>), L<http://www.cygwin.com/> |
c997b287 GS |
1041 | |
1042 | =item * | |
1043 | ||
1044 | The U/WIN environment for Win32, | |
500f1b69 | 1045 | L<http://www.research.att.com/sw/tools/uwin/> |
c997b287 | 1046 | |
cea6626f | 1047 | =item * |
d1e3b762 | 1048 | |
cea6626f | 1049 | Build instructions for OS/2, L<perlos2> |
d1e3b762 | 1050 | |
e41182b5 GS |
1051 | =back |
1052 | ||
e41182b5 GS |
1053 | =head2 VMS |
1054 | ||
2c044526 | 1055 | Perl on VMS is discussed in L<perlvms> in the Perl distribution. |
016930a6 JM |
1056 | |
1057 | The official name of VMS as of this writing is OpenVMS. | |
1058 | ||
e41182b5 GS |
1059 | Interacting with Perl from the Digital Command Language (DCL) shell |
1060 | often requires a different set of quotation marks than Unix shells do. | |
1061 | For example: | |
1062 | ||
1063 | $ perl -e "print ""Hello, world.\n""" | |
1064 | Hello, world. | |
1065 | ||
2c044526 | 1066 | There are several ways to wrap your Perl scripts in DCL F<.COM> files, if |
e41182b5 GS |
1067 | you are so inclined. For example: |
1068 | ||
1069 | $ write sys$output "Hello from DCL!" | |
1070 | $ if p1 .eqs. "" | |
1071 | $ then perl -x 'f$environment("PROCEDURE") | |
1072 | $ else perl -x - 'p1 'p2 'p3 'p4 'p5 'p6 'p7 'p8 | |
1073 | $ deck/dollars="__END__" | |
1074 | #!/usr/bin/perl | |
1075 | ||
1076 | print "Hello from Perl!\n"; | |
1077 | ||
1078 | __END__ | |
1079 | $ endif | |
1080 | ||
1081 | Do take care with C<$ ASSIGN/nolog/user SYS$COMMAND: SYS$INPUT> if your | |
2c044526 | 1082 | Perl-in-DCL script expects to do things like C<< $read = <STDIN>; >>. |
e41182b5 | 1083 | |
9e861032 CB |
1084 | The VMS operating system has two filesystems, designated by their |
1085 | on-disk structure (ODS) level: ODS-2 and its successor ODS-5. The | |
1086 | initial port of Perl to VMS pre-dates ODS-5, but all current testing and | |
1087 | development assumes ODS-5 and its capabilities, including case | |
1088 | preservation, extended characters in filespecs, and names up to 8192 | |
1089 | bytes long. | |
016930a6 | 1090 | |
9e861032 CB |
1091 | Perl on VMS can accept either VMS- or Unix-style file |
1092 | specifications as in either of the following: | |
1089a9e3 | 1093 | |
9e861032 CB |
1094 | $ perl -ne "print if /perl_setup/i" SYS$LOGIN:LOGIN.COM |
1095 | $ perl -ne "print if /perl_setup/i" /sys$login/login.com | |
1089a9e3 | 1096 | |
9e861032 | 1097 | but not a mixture of both as in: |
1089a9e3 | 1098 | |
9e861032 CB |
1099 | $ perl -ne "print if /perl_setup/i" sys$login:/login.com |
1100 | Can't open sys$login:/login.com: file specification syntax error | |
e41182b5 | 1101 | |
9e861032 CB |
1102 | In general, the easiest path to portability is always to specify |
1103 | filenames in Unix format unless they will need to be processed by native | |
1104 | commands or utilities. Because of this latter consideration, the | |
83a46a63 | 1105 | L<File::Spec> module by default returns native format specifications |
9e861032 CB |
1106 | regardless of input format. This default may be reversed so that |
1107 | filenames are always reported in Unix format by specifying the | |
1108 | C<DECC$FILENAME_UNIX_REPORT> feature logical in the environment. | |
1109 | ||
1110 | The file type, or extension, is always present in a VMS-format file | |
1111 | specification even if it's zero-length. This means that, by default, | |
83a46a63 LM |
1112 | L<C<readdir>|perlfunc/readdir DIRHANDLE> will return a trailing dot on a |
1113 | file with no extension, so where you would see C<"a"> on Unix you'll see | |
1114 | C<"a."> on VMS. However, the trailing dot may be suppressed by enabling | |
1115 | the C<DECC$READDIR_DROPDOTNOTYPE> feature in the environment (see the CRTL | |
9e861032 | 1116 | documentation on feature logical names). |
e41182b5 | 1117 | |
5e12dbfa | 1118 | What C<\n> represents depends on the type of file opened. It usually |
2890cc8c CBW |
1119 | represents C<\012> but it could also be C<\015>, C<\012>, C<\015\012>, |
1120 | C<\000>, C<\040>, or nothing depending on the file organization and | |
83a46a63 LM |
1121 | record format. The L<C<VMS::Stdio>|VMS::Stdio> module provides access to |
1122 | the special C<fopen()> requirements of files with unusual attributes on | |
1123 | VMS. | |
e41182b5 | 1124 | |
83a46a63 LM |
1125 | The value of L<C<$^O>|perlvar/$^O> on OpenVMS is "VMS". To determine the |
1126 | architecture that you are running on refer to | |
1127 | L<C<$Config{archname}>|Config/C<archname>>. | |
016930a6 | 1128 | |
b7df3edc GS |
1129 | On VMS, perl determines the UTC offset from the C<SYS$TIMEZONE_DIFFERENTIAL> |
1130 | logical name. Although the VMS epoch began at 17-NOV-1858 00:00:00.00, | |
83a46a63 LM |
1131 | calls to L<C<localtime>|perlfunc/localtime EXPR> are adjusted to count |
1132 | offsets from 01-JAN-1970 00:00:00.00, just like Unix. | |
6ab3f9cb | 1133 | |
e41182b5 GS |
1134 | Also see: |
1135 | ||
1136 | =over 4 | |
1137 | ||
c997b287 GS |
1138 | =item * |
1139 | ||
96090e4f | 1140 | F<README.vms> (installed as F<README_vms>), L<perlvms> |
c997b287 GS |
1141 | |
1142 | =item * | |
1143 | ||
1089a9e3 | 1144 | vmsperl list, vmsperl-subscribe@perl.org |
e41182b5 | 1145 | |
c997b287 | 1146 | =item * |
e41182b5 | 1147 | |
500f1b69 | 1148 | vmsperl on the web, L<http://www.sidhe.org/vmsperl/index.html> |
e41182b5 | 1149 | |
9e861032 CB |
1150 | =item * |
1151 | ||
1152 | VMS Software Inc. web site, L<http://www.vmssoftware.com> | |
1153 | ||
e41182b5 GS |
1154 | =back |
1155 | ||
495c5fdc PG |
1156 | =head2 VOS |
1157 | ||
10fb90aa | 1158 | Perl on VOS (also known as OpenVOS) is discussed in F<README.vos> |
2c044526 | 1159 | in the Perl distribution (installed as L<perlvos>). Perl on VOS |
10fb90aa PG |
1160 | can accept either VOS- or Unix-style file specifications as in |
1161 | either of the following: | |
495c5fdc | 1162 | |
ea8b8ad2 VP |
1163 | $ perl -ne "print if /perl_setup/i" >system>notices |
1164 | $ perl -ne "print if /perl_setup/i" /system/notices | |
495c5fdc PG |
1165 | |
1166 | or even a mixture of both as in: | |
1167 | ||
ea8b8ad2 | 1168 | $ perl -ne "print if /perl_setup/i" >system/notices |
495c5fdc | 1169 | |
b7df3edc | 1170 | Even though VOS allows the slash character to appear in object |
495c5fdc | 1171 | names, because the VOS port of Perl interprets it as a pathname |
10fb90aa PG |
1172 | delimiting character, VOS files, directories, or links whose |
1173 | names contain a slash character cannot be processed. Such files | |
1174 | must be renamed before they can be processed by Perl. | |
1175 | ||
1176 | Older releases of VOS (prior to OpenVOS Release 17.0) limit file | |
1177 | names to 32 or fewer characters, prohibit file names from | |
1178 | starting with a C<-> character, and prohibit file names from | |
83a46a63 | 1179 | containing C< > (space) or any character from the set C<< !#%&'()*;<=>? >>. |
10fb90aa PG |
1180 | |
1181 | Newer releases of VOS (OpenVOS Release 17.0 or later) support a | |
1182 | feature known as extended names. On these releases, file names | |
1183 | can contain up to 255 characters, are prohibited from starting | |
1184 | with a C<-> character, and the set of prohibited characters is | |
83a46a63 | 1185 | reduced to C<< #%*<>? >>. There are |
c69ca1d4 | 1186 | restrictions involving spaces and apostrophes: these characters |
10fb90aa PG |
1187 | must not begin or end a name, nor can they immediately precede or |
1188 | follow a period. Additionally, a space must not immediately | |
1189 | precede another space or hyphen. Specifically, the following | |
1190 | character combinations are prohibited: space-space, | |
1191 | space-hyphen, period-space, space-period, period-apostrophe, | |
1192 | apostrophe-period, leading or trailing space, and leading or | |
1193 | trailing apostrophe. Although an extended file name is limited | |
1194 | to 255 characters, a path name is still limited to 256 | |
1195 | characters. | |
1196 | ||
83a46a63 LM |
1197 | The value of L<C<$^O>|perlvar/$^O> on VOS is "vos". To determine the |
1198 | architecture that you are running on refer to | |
1199 | L<C<$Config{archname}>|Config/C<archname>>. | |
495c5fdc | 1200 | |
495c5fdc PG |
1201 | Also see: |
1202 | ||
1203 | =over 4 | |
1204 | ||
c997b287 | 1205 | =item * |
495c5fdc | 1206 | |
cc07ed0b | 1207 | F<README.vos> (installed as L<perlvos>) |
c997b287 GS |
1208 | |
1209 | =item * | |
1210 | ||
1211 | The VOS mailing list. | |
495c5fdc | 1212 | |
7d4dfb6d PG |
1213 | There is no specific mailing list for Perl on VOS. You can contact |
1214 | the Stratus Technologies Customer Assistance Center (CAC) for your | |
1215 | region, or you can use the contact information located in the | |
1216 | distribution files on the Stratus Anonymous FTP site. | |
495c5fdc | 1217 | |
c997b287 GS |
1218 | =item * |
1219 | ||
7d4dfb6d PG |
1220 | Stratus Technologies on the web at L<http://www.stratus.com> |
1221 | ||
1222 | =item * | |
1223 | ||
1224 | VOS Open-Source Software on the web at L<http://ftp.stratus.com/pub/vos/vos.html> | |
495c5fdc PG |
1225 | |
1226 | =back | |
1227 | ||
e41182b5 GS |
1228 | =head2 EBCDIC Platforms |
1229 | ||
6eb53dea KW |
1230 | v5.22 core Perl runs on z/OS (formerly OS/390). Theoretically it could |
1231 | run on the successors of OS/400 on AS/400 minicomputers as well as | |
1232 | VM/ESA, and BS2000 for S/390 Mainframes. Such computers use EBCDIC | |
83a46a63 LM |
1233 | character sets internally (usually Character Code Set ID 0037 for OS/400 |
1234 | and either 1047 or POSIX-BC for S/390 systems). | |
6eb53dea KW |
1235 | |
1236 | The rest of this section may need updating, but we don't know what it | |
1237 | should say. Please email comments to | |
1238 | L<perlbug@perl.org|mailto:perlbug@perl.org>. | |
1239 | ||
1240 | On the mainframe Perl currently works under the "Unix system | |
0cc436d0 | 1241 | services for OS/390" (formerly known as OpenEdition), VM/ESA OpenEdition, or |
2c044526 | 1242 | the BS200 POSIX-BC system (BS2000 is supported in Perl 5.6 and greater). |
522b859a | 1243 | See L<perlos390> for details. Note that for OS/400 there is also a port of |
1bcbdd38 | 1244 | Perl 5.8.1/5.10.0 or later to the PASE which is ASCII-based (as opposed to |
2890cc8c | 1245 | ILE which is EBCDIC-based), see L<perlos400>. |
e41182b5 | 1246 | |
7c5ffed3 JH |
1247 | As of R2.5 of USS for OS/390 and Version 2.3 of VM/ESA these Unix |
1248 | sub-systems do not support the C<#!> shebang trick for script invocation. | |
2c044526 | 1249 | Hence, on OS/390 and VM/ESA Perl scripts can be executed with a header |
7c5ffed3 | 1250 | similar to the following simple script: |
e41182b5 GS |
1251 | |
1252 | : # use perl | |
1253 | eval 'exec /usr/local/bin/perl -S $0 ${1+"$@"}' | |
1254 | if 0; | |
1255 | #!/usr/local/bin/perl # just a comment really | |
1256 | ||
1257 | print "Hello from perl!\n"; | |
1258 | ||
d1e3b762 | 1259 | OS/390 will support the C<#!> shebang trick in release 2.8 and beyond. |
83a46a63 LM |
1260 | Calls to L<C<system>|perlfunc/system LIST> and backticks can use POSIX |
1261 | shell syntax on all S/390 systems. | |
d1e3b762 | 1262 | |
b7df3edc | 1263 | On the AS/400, if PERL5 is in your library list, you may need |
2c044526 | 1264 | to wrap your Perl scripts in a CL procedure to invoke them like so: |
6ab3f9cb GS |
1265 | |
1266 | BEGIN | |
1267 | CALL PGM(PERL5/PERL) PARM('/QOpenSys/hello.pl') | |
1268 | ENDPGM | |
1269 | ||
2c044526 | 1270 | This will invoke the Perl script F<hello.pl> in the root of the |
83a46a63 LM |
1271 | QOpenSys file system. On the AS/400 calls to |
1272 | L<C<system>|perlfunc/system LIST> or backticks must use CL syntax. | |
6ab3f9cb | 1273 | |
e41182b5 | 1274 | On these platforms, bear in mind that the EBCDIC character set may have |
83a46a63 LM |
1275 | an effect on what happens with some Perl functions (such as |
1276 | L<C<chr>|perlfunc/chr NUMBER>, L<C<pack>|perlfunc/pack TEMPLATE,LIST>, | |
1277 | L<C<print>|perlfunc/print FILEHANDLE LIST>, | |
1278 | L<C<printf>|perlfunc/printf FILEHANDLE FORMAT, LIST>, | |
1279 | L<C<ord>|perlfunc/ord EXPR>, L<C<sort>|perlfunc/sort SUBNAME LIST>, | |
1280 | L<C<sprintf>|perlfunc/sprintf FORMAT, LIST>, | |
1281 | L<C<unpack>|perlfunc/unpack TEMPLATE,EXPR>), as | |
1282 | well as bit-fiddling with ASCII constants using operators like | |
1283 | L<C<^>, C<&> and C<|>|perlop/Bitwise String Operators>, not to mention | |
1284 | dealing with socket interfaces to ASCII computers (see L</"Newlines">). | |
e41182b5 | 1285 | |
b7df3edc GS |
1286 | Fortunately, most web servers for the mainframe will correctly |
1287 | translate the C<\n> in the following statement to its ASCII equivalent | |
2c044526 | 1288 | (C<\r> is the same under both Unix and z/OS): |
e41182b5 GS |
1289 | |
1290 | print "Content-type: text/html\r\n\r\n"; | |
1291 | ||
83a46a63 | 1292 | The values of L<C<$^O>|perlvar/$^O> on some of these platforms include: |
e41182b5 | 1293 | |
83a46a63 | 1294 | uname $^O $Config{archname} |
d1e3b762 GS |
1295 | -------------------------------------------- |
1296 | OS/390 os390 os390 | |
1297 | OS400 os400 os400 | |
1298 | POSIX-BC posix-bc BS2000-posix-bc | |
3c075c7d | 1299 | |
e41182b5 GS |
1300 | Some simple tricks for determining if you are running on an EBCDIC |
1301 | platform could include any of the following (perhaps all): | |
1302 | ||
83a46a63 | 1303 | if ("\t" eq "\005") { print "EBCDIC may be spoken here!\n"; } |
e41182b5 GS |
1304 | |
1305 | if (ord('A') == 193) { print "EBCDIC may be spoken here!\n"; } | |
1306 | ||
1307 | if (chr(169) eq 'z') { print "EBCDIC may be spoken here!\n"; } | |
1308 | ||
b7df3edc | 1309 | One thing you may not want to rely on is the EBCDIC encoding |
0a47030a GS |
1310 | of punctuation characters since these may differ from code page to code |
1311 | page (and once your module or script is rumoured to work with EBCDIC, | |
1312 | folks will want it to work with all EBCDIC character sets). | |
e41182b5 GS |
1313 | |
1314 | Also see: | |
1315 | ||
1316 | =over 4 | |
1317 | ||
c997b287 GS |
1318 | =item * |
1319 | ||
6eb53dea | 1320 | L<perlos390>, L<perlos400>, L<perlbs2000>, L<perlebcdic>. |
c997b287 GS |
1321 | |
1322 | =item * | |
e41182b5 GS |
1323 | |
1324 | The perl-mvs@perl.org list is for discussion of porting issues as well as | |
1325 | general usage issues for all EBCDIC Perls. Send a message body of | |
1326 | "subscribe perl-mvs" to majordomo@perl.org. | |
1327 | ||
7ee27b7c | 1328 | =item * |
c997b287 GS |
1329 | |
1330 | AS/400 Perl information at | |
500f1b69 | 1331 | L<http://as400.rochester.ibm.com/> |
d1e3b762 | 1332 | as well as on CPAN in the F<ports/> directory. |
e41182b5 GS |
1333 | |
1334 | =back | |
1335 | ||
b8099c3d CN |
1336 | =head2 Acorn RISC OS |
1337 | ||
b7df3edc | 1338 | Because Acorns use ASCII with newlines (C<\n>) in text files as C<\012> like |
2890cc8c | 1339 | Unix, and because Unix filename emulation is turned on by default, |
b7df3edc | 1340 | most simple scripts will probably work "out of the box". The native |
6ab3f9cb | 1341 | filesystem is modular, and individual filesystems are free to be |
0a47030a | 1342 | case-sensitive or insensitive, and are usually case-preserving. Some |
b7df3edc | 1343 | native filesystems have name length limits, which file and directory |
6ab3f9cb GS |
1344 | names are silently truncated to fit. Scripts should be aware that the |
1345 | standard filesystem currently has a name length limit of B<10> | |
1346 | characters, with up to 77 items in a directory, but other filesystems | |
0a47030a | 1347 | may not impose such limitations. |
b8099c3d CN |
1348 | |
1349 | Native filenames are of the form | |
1350 | ||
6ab3f9cb | 1351 | Filesystem#Special_Field::DiskName.$.Directory.Directory.File |
dd9f0070 | 1352 | |
b8099c3d CN |
1353 | where |
1354 | ||
1355 | Special_Field is not usually present, but may contain . and $ . | |
1356 | Filesystem =~ m|[A-Za-z0-9_]| | |
1357 | DsicName =~ m|[A-Za-z0-9_/]| | |
1358 | $ represents the root directory | |
1359 | . is the path separator | |
1360 | @ is the current directory (per filesystem but machine global) | |
1361 | ^ is the parent directory | |
1362 | Directory and File =~ m|[^\0- "\.\$\%\&:\@\\^\|\177]+| | |
1363 | ||
83a46a63 | 1364 | The default filename translation is roughly C<tr|/.|./|>, swapping dots |
dabde021 | 1365 | and slashes. |
b8099c3d | 1366 | |
6ab3f9cb | 1367 | Note that C<"ADFS::HardDisk.$.File" ne 'ADFS::HardDisk.$.File'> and that |
0a47030a | 1368 | the second stage of C<$> interpolation in regular expressions will fall |
83a46a63 | 1369 | foul of the L<C<$.>|perlvar/$.> variable if scripts are not careful. |
0a47030a GS |
1370 | |
1371 | Logical paths specified by system variables containing comma-separated | |
b7df3edc | 1372 | search lists are also allowed; hence C<System:Modules> is a valid |
0a47030a | 1373 | filename, and the filesystem will prefix C<Modules> with each section of |
6ab3f9cb | 1374 | C<System$Path> until a name is made that points to an object on disk. |
b7df3edc | 1375 | Writing to a new file C<System:Modules> would be allowed only if |
0a47030a GS |
1376 | C<System$Path> contains a single item list. The filesystem will also |
1377 | expand system variables in filenames if enclosed in angle brackets, so | |
c47ff5f1 | 1378 | C<< <System$Dir>.Modules >> would look for the file |
0a47030a | 1379 | S<C<$ENV{'System$Dir'} . 'Modules'>>. The obvious implication of this is |
83a46a63 LM |
1380 | that B<fully qualified filenames can start with C<< <> >>> and the |
1381 | three-argument form of L<C<open>|perlfunc/open FILEHANDLE,EXPR> should | |
1382 | always be used. | |
b8099c3d CN |
1383 | |
1384 | Because C<.> was in use as a directory separator and filenames could not | |
1385 | be assumed to be unique after 10 characters, Acorn implemented the C | |
1386 | compiler to strip the trailing C<.c> C<.h> C<.s> and C<.o> suffix from | |
1387 | filenames specified in source code and store the respective files in | |
b7df3edc | 1388 | subdirectories named after the suffix. Hence files are translated: |
b8099c3d CN |
1389 | |
1390 | foo.h h.foo | |
1391 | C:foo.h C:h.foo (logical path variable) | |
1392 | sys/os.h sys.h.os (C compiler groks Unix-speak) | |
1393 | 10charname.c c.10charname | |
1394 | 10charname.o o.10charname | |
1395 | 11charname_.c c.11charname (assuming filesystem truncates at 10) | |
1396 | ||
1397 | The Unix emulation library's translation of filenames to native assumes | |
b7df3edc GS |
1398 | that this sort of translation is required, and it allows a user-defined list |
1399 | of known suffixes that it will transpose in this fashion. This may | |
92863ac5 | 1400 | seem transparent, but consider that with these rules F<foo/bar/baz.h> |
83a46a63 LM |
1401 | and F<foo/bar/h/baz> both map to F<foo.bar.h.baz>, and that |
1402 | L<C<readdir>|perlfunc/readdir DIRHANDLE> and L<C<glob>|perlfunc/glob EXPR> | |
1403 | cannot and do not attempt to emulate the reverse mapping. Other | |
6ab3f9cb | 1404 | C<.>'s in filenames are translated to C</>. |
0a47030a | 1405 | |
83a46a63 LM |
1406 | As implied above, the environment accessed through |
1407 | L<C<%ENV>|perlvar/%ENV> is global, and the convention is that program | |
1408 | specific environment variables are of the form C<Program$Name>. | |
1409 | Each filesystem maintains a current directory, | |
6ab3f9cb | 1410 | and the current filesystem's current directory is the B<global> current |
b7df3edc GS |
1411 | directory. Consequently, sociable programs don't change the current |
1412 | directory but rely on full pathnames, and programs (and Makefiles) cannot | |
0a47030a GS |
1413 | assume that they can spawn a child process which can change the current |
1414 | directory without affecting its parent (and everyone else for that | |
1415 | matter). | |
1416 | ||
2890cc8c | 1417 | Because native operating system filehandles are global and are currently |
b7df3edc | 1418 | allocated down from 255, with 0 being a reserved value, the Unix emulation |
0a47030a GS |
1419 | library emulates Unix filehandles. Consequently, you can't rely on |
1420 | passing C<STDIN>, C<STDOUT>, or C<STDERR> to your children. | |
1421 | ||
1422 | The desire of users to express filenames of the form | |
c47ff5f1 | 1423 | C<< <Foo$Dir>.Bar >> on the command line unquoted causes problems, |
83a46a63 LM |
1424 | too: L<C<``>|perlop/C<qxE<sol>I<STRING>E<sol>>> command output capture has |
1425 | to perform a guessing game. It assumes that a string C<< <[^<>]+\$[^<>]> >> | |
1426 | is a reference to an environment variable, whereas anything else involving | |
c47ff5f1 | 1427 | C<< < >> or C<< > >> is redirection, and generally manages to be 99% |
0a47030a GS |
1428 | right. Of course, the problem remains that scripts cannot rely on any |
1429 | Unix tools being available, or that any tools found have Unix-like command | |
1430 | line arguments. | |
1431 | ||
b7df3edc GS |
1432 | Extensions and XS are, in theory, buildable by anyone using free |
1433 | tools. In practice, many don't, as users of the Acorn platform are | |
1434 | used to binary distributions. MakeMaker does run, but no available | |
1435 | make currently copes with MakeMaker's makefiles; even if and when | |
1436 | this should be fixed, the lack of a Unix-like shell will cause | |
83a46a63 LM |
1437 | problems with makefile rules, especially lines of the form |
1438 | C<cd sdbm && make all>, and anything using quoting. | |
b8099c3d | 1439 | |
83a46a63 LM |
1440 | S<"RISC OS"> is the proper name for the operating system, but the value |
1441 | in L<C<$^O>|perlvar/$^O> is "riscos" (because we don't like shouting). | |
b8099c3d | 1442 | |
e41182b5 GS |
1443 | =head2 Other perls |
1444 | ||
b7df3edc | 1445 | Perl has been ported to many platforms that do not fit into any of |
b6c36746 | 1446 | the categories listed above. Some, such as AmigaOS, |
cd86ed9d JV |
1447 | QNX, Plan 9, and VOS, have been well-integrated into the standard |
1448 | Perl source code kit. You may need to see the F<ports/> directory | |
1449 | on CPAN for information, and possibly binaries, for the likes of: | |
1450 | aos, Atari ST, lynxos, riscos, Novell Netware, Tandem Guardian, | |
1451 | I<etc.> (Yes, we know that some of these OSes may fall under the | |
1452 | Unix category, but we are not a standards body.) | |
e41182b5 | 1453 | |
83a46a63 LM |
1454 | Some approximate operating system names and their L<C<$^O>|perlvar/$^O> |
1455 | values in the "OTHER" category include: | |
d1e3b762 | 1456 | |
83a46a63 | 1457 | OS $^O $Config{archname} |
d1e3b762 GS |
1458 | ------------------------------------------ |
1459 | Amiga DOS amigaos m68k-amigos | |
d1e3b762 | 1460 | |
e41182b5 GS |
1461 | See also: |
1462 | ||
1463 | =over 4 | |
1464 | ||
c997b287 GS |
1465 | =item * |
1466 | ||
1467 | Amiga, F<README.amiga> (installed as L<perlamiga>). | |
1468 | ||
1469 | =item * | |
d1e3b762 | 1470 | |
6ab3f9cb | 1471 | A free perl5-based PERL.NLM for Novell Netware is available in |
500f1b69 | 1472 | precompiled binary and source code form from L<http://www.novell.com/> |
6ab3f9cb | 1473 | as well as from CPAN. |
e41182b5 | 1474 | |
13a2d996 | 1475 | =item * |
c997b287 | 1476 | |
e6f03d26 | 1477 | S<Plan 9>, F<README.plan9> |
d1e3b762 | 1478 | |
e41182b5 GS |
1479 | =back |
1480 | ||
e41182b5 GS |
1481 | =head1 FUNCTION IMPLEMENTATIONS |
1482 | ||
b7df3edc GS |
1483 | Listed below are functions that are either completely unimplemented |
1484 | or else have been implemented differently on various platforms. | |
d23c3b6f | 1485 | Preceding each description will be, in parentheses, a list of |
b7df3edc | 1486 | platforms that the description applies to. |
e41182b5 | 1487 | |
b7df3edc GS |
1488 | The list may well be incomplete, or even wrong in some places. When |
1489 | in doubt, consult the platform-specific README files in the Perl | |
1490 | source distribution, and any other documentation resources accompanying | |
1491 | a given port. | |
e41182b5 | 1492 | |
0a47030a | 1493 | Be aware, moreover, that even among Unix-ish systems there are variations. |
e41182b5 | 1494 | |
83a46a63 LM |
1495 | For many functions, you can also query L<C<%Config>|Config/DESCRIPTION>, |
1496 | exported by default from the L<C<Config>|Config> module. For example, to | |
1497 | check whether the platform has the L<C<lstat>|perlfunc/lstat FILEHANDLE> | |
1498 | call, check L<C<$Config{d_lstat}>|Config/C<d_lstat>>. See L<Config> for a | |
1499 | full description of available variables. | |
e41182b5 GS |
1500 | |
1501 | =head2 Alphabetical Listing of Perl Functions | |
1502 | ||
1503 | =over 8 | |
1504 | ||
e41182b5 GS |
1505 | =item -X |
1506 | ||
d23c3b6f | 1507 | (Win32) |
038ae9a4 SH |
1508 | C<-w> only inspects the read-only file attribute (FILE_ATTRIBUTE_READONLY), |
1509 | which determines whether the directory can be deleted, not whether it can | |
1510 | be written to. Directories always have read and write access unless denied | |
d23c3b6f | 1511 | by discretionary access control lists (DACLs). |
038ae9a4 | 1512 | |
d23c3b6f | 1513 | (VMS) |
b7df3edc | 1514 | C<-r>, C<-w>, C<-x>, and C<-o> tell whether the file is accessible, |
d23c3b6f | 1515 | which may not reflect UIC-based file protections. |
e41182b5 | 1516 | |
d23c3b6f | 1517 | (S<RISC OS>) |
b8099c3d CN |
1518 | C<-s> by name on an open file will return the space reserved on disk, |
1519 | rather than the current extent. C<-s> on an open filehandle returns the | |
d23c3b6f | 1520 | current size. |
b8099c3d | 1521 | |
d23c3b6f | 1522 | (Win32, VMS, S<RISC OS>) |
e41182b5 | 1523 | C<-R>, C<-W>, C<-X>, C<-O> are indistinguishable from C<-r>, C<-w>, |
d23c3b6f | 1524 | C<-x>, C<-o>. |
e41182b5 | 1525 | |
b8099c3d | 1526 | (Win32, VMS, S<RISC OS>) |
d23c3b6f | 1527 | C<-g>, C<-k>, C<-l>, C<-u>, C<-A> are not particularly meaningful. |
e41182b5 | 1528 | |
d23c3b6f LM |
1529 | (VMS, S<RISC OS>) |
1530 | C<-p> is not particularly meaningful. | |
287a962e | 1531 | |
e41182b5 | 1532 | (VMS) |
d23c3b6f | 1533 | C<-d> is true if passed a device spec without an explicit directory. |
e41182b5 | 1534 | |
d23c3b6f | 1535 | (Win32) |
e41182b5 | 1536 | C<-x> (or C<-X>) determine if a file ends in one of the executable |
d23c3b6f | 1537 | suffixes. C<-S> is meaningless. |
e41182b5 | 1538 | |
b8099c3d | 1539 | (S<RISC OS>) |
d23c3b6f | 1540 | C<-x> (or C<-X>) determine if a file has an executable file type. |
b8099c3d | 1541 | |
aca72608 JD |
1542 | =item alarm |
1543 | ||
d23c3b6f | 1544 | (Win32) |
aca72608 JD |
1545 | Emulated using timers that must be explicitly polled whenever Perl |
1546 | wants to dispatch "safe signals" and therefore cannot interrupt | |
d23c3b6f | 1547 | blocking system calls. |
aca72608 | 1548 | |
47cd99a4 | 1549 | =item atan2 |
519bc777 | 1550 | |
d23c3b6f | 1551 | (Tru64, HP-UX 10.20) |
519bc777 | 1552 | Due to issues with various CPUs, math libraries, compilers, and standards, |
83a46a63 | 1553 | results for C<atan2> may vary depending on any combination of the above. |
519bc777 | 1554 | Perl attempts to conform to the Open Group/IEEE standards for the results |
83a46a63 | 1555 | returned from C<atan2>, but cannot force the issue if the system Perl is |
d23c3b6f | 1556 | run on does not allow it. |
519bc777 | 1557 | |
83a46a63 | 1558 | The current version of the standards for C<atan2> is available at |
519bc777 RGS |
1559 | L<http://www.opengroup.org/onlinepubs/009695399/functions/atan2.html>. |
1560 | ||
47cd99a4 | 1561 | =item binmode |
e41182b5 | 1562 | |
d23c3b6f LM |
1563 | (S<RISC OS>) |
1564 | Meaningless. | |
e41182b5 | 1565 | |
d23c3b6f | 1566 | (VMS) |
e41182b5 GS |
1567 | Reopens file and restores pointer; if function fails, underlying |
1568 | filehandle may be closed, or pointer may be in a different position. | |
e41182b5 | 1569 | |
d23c3b6f | 1570 | (Win32) |
83a46a63 | 1571 | The value returned by L<C<tell>|perlfunc/tell FILEHANDLE> may be affected |
d23c3b6f | 1572 | after the call, and the filehandle may be flushed. |
e41182b5 | 1573 | |
47cd99a4 | 1574 | =item chmod |
e41182b5 | 1575 | |
d23c3b6f | 1576 | (Win32) |
83a46a63 | 1577 | Only good for changing "owner" read-write access; "group" and "other" |
d23c3b6f | 1578 | bits are meaningless. |
e41182b5 | 1579 | |
d23c3b6f LM |
1580 | (S<RISC OS>) |
1581 | Only good for changing "owner" and "other" read-write access. | |
b8099c3d | 1582 | |
d23c3b6f LM |
1583 | (VOS) |
1584 | Access permissions are mapped onto VOS access-control list changes. | |
495c5fdc | 1585 | |
d23c3b6f | 1586 | (Cygwin) |
83a46a63 | 1587 | The actual permissions set depend on the value of the C<CYGWIN> variable |
d23c3b6f | 1588 | in the SYSTEM environment settings. |
4e51f8e4 | 1589 | |
d23c3b6f | 1590 | (Android) |
2c044526 | 1591 | Setting the exec bit on some locations (generally F</sdcard>) will return true |
d23c3b6f | 1592 | but not actually set the bit. |
43b08d74 | 1593 | |
669d6ad8 CB |
1594 | (VMS) |
1595 | A mode argument of zero sets permissions to the user's default permission mask | |
1596 | rather than disabling all permissions. | |
1597 | ||
47cd99a4 | 1598 | =item chown |
e41182b5 | 1599 | |
d23c3b6f LM |
1600 | (S<Plan 9>, S<RISC OS>) |
1601 | Not implemented. | |
e41182b5 | 1602 | |
d23c3b6f LM |
1603 | (Win32) |
1604 | Does nothing, but won't fail. | |
e41182b5 | 1605 | |
d23c3b6f LM |
1606 | (VOS) |
1607 | A little funky, because VOS's notion of ownership is a little funky. | |
3fd80bd6 | 1608 | |
e41182b5 GS |
1609 | =item chroot |
1610 | ||
d23c3b6f LM |
1611 | (Win32, VMS, S<Plan 9>, S<RISC OS>, VOS) |
1612 | Not implemented. | |
e41182b5 | 1613 | |
47cd99a4 | 1614 | =item crypt |
e41182b5 | 1615 | |
d23c3b6f | 1616 | (Win32) |
e41182b5 | 1617 | May not be available if library or source was not provided when building |
d23c3b6f | 1618 | perl. |
e41182b5 | 1619 | |
d23c3b6f LM |
1620 | (Android) |
1621 | Not implemented. | |
43b08d74 | 1622 | |
47cd99a4 | 1623 | =item dbmclose |
e41182b5 | 1624 | |
d23c3b6f LM |
1625 | (VMS, S<Plan 9>, VOS) |
1626 | Not implemented. | |
e41182b5 | 1627 | |
47cd99a4 | 1628 | =item dbmopen |
e41182b5 | 1629 | |
d23c3b6f LM |
1630 | (VMS, S<Plan 9>, VOS) |
1631 | Not implemented. | |
e41182b5 | 1632 | |
47cd99a4 | 1633 | =item dump |
e41182b5 | 1634 | |
d23c3b6f LM |
1635 | (S<RISC OS>) |
1636 | Not useful. | |
e41182b5 | 1637 | |
d23c3b6f LM |
1638 | (Cygwin, Win32) |
1639 | Not supported. | |
e41182b5 | 1640 | |
d23c3b6f LM |
1641 | (VMS) |
1642 | Invokes VMS debugger. | |
e41182b5 | 1643 | |
47cd99a4 | 1644 | =item exec |
e41182b5 | 1645 | |
d23c3b6f | 1646 | (Win32) |
94d4006a | 1647 | C<exec LIST> without the use of indirect object syntax (C<exec PROGRAM LIST>) |
d23c3b6f | 1648 | may fall back to trying the shell if the first C<spawn()> fails. |
94d4006a | 1649 | |
0f897271 | 1650 | (SunOS, Solaris, HP-UX) |
d23c3b6f | 1651 | Does not automatically flush output handles on some platforms. |
0f897271 | 1652 | |
d23c3b6f LM |
1653 | (Symbian OS) |
1654 | Not supported. | |
af8bb25a | 1655 | |
fe12c0e8 MS |
1656 | =item exit |
1657 | ||
d23c3b6f | 1658 | (VMS) |
83a46a63 | 1659 | Emulates Unix C<exit> (which considers C<exit 1> to indicate an error) by |
2c044526 | 1660 | mapping the C<1> to C<SS$_ABORT> (C<44>). This behavior may be overridden |
83a46a63 LM |
1661 | with the pragma L<C<use vmsish 'exit'>|vmsish/C<vmsish exit>>. As with |
1662 | the CRTL's C<exit()> function, C<exit 0> is also mapped to an exit status | |
1663 | of C<SS$_NORMAL> (C<1>); this mapping cannot be overridden. Any other | |
1664 | argument to C<exit> | |
016930a6 JM |
1665 | is used directly as Perl's exit status. On VMS, unless the future |
1666 | POSIX_EXIT mode is enabled, the exit code should always be a valid | |
1667 | VMS exit code and not a generic number. When the POSIX_EXIT mode is | |
1668 | enabled, a generic number will be encoded in a method compatible with | |
1669 | the C library _POSIX_EXIT macro so that it can be decoded by other | |
d23c3b6f | 1670 | programs, particularly ones written in C, like the GNV package. |
fe12c0e8 | 1671 | |
d23c3b6f | 1672 | (Solaris) |
83a46a63 LM |
1673 | C<exit> resets file pointers, which is a problem when called |
1674 | from a child process (created by L<C<fork>|perlfunc/fork>) in | |
1675 | L<C<BEGIN>|perlmod/BEGIN, UNITCHECK, CHECK, INIT and END>. | |
d23c3b6f | 1676 | A workaround is to use L<C<POSIX::_exit>|POSIX/C<_exit>>. |
bef2c191 RB |
1677 | |
1678 | exit unless $Config{archname} =~ /\bsolaris\b/; | |
83a46a63 LM |
1679 | require POSIX; |
1680 | POSIX::_exit(0); | |
bef2c191 | 1681 | |
47cd99a4 | 1682 | =item fcntl |
e41182b5 | 1683 | |
d23c3b6f LM |
1684 | (Win32) |
1685 | Not implemented. | |
6a065175 | 1686 | |
d23c3b6f LM |
1687 | (VMS) |
1688 | Some functions available based on the version of VMS. | |
e41182b5 | 1689 | |
47cd99a4 | 1690 | =item flock |
e41182b5 | 1691 | |
d23c3b6f LM |
1692 | (VMS, S<RISC OS>, VOS) |
1693 | Not implemented. | |
e41182b5 | 1694 | |
e41182b5 GS |
1695 | =item fork |
1696 | ||
d23c3b6f LM |
1697 | (AmigaOS, S<RISC OS>, VMS) |
1698 | Not implemented. | |
0f897271 | 1699 | |
d23c3b6f LM |
1700 | (Win32) |
1701 | Emulated using multiple interpreters. See L<perlfork>. | |
0f897271 | 1702 | |
0f897271 | 1703 | (SunOS, Solaris, HP-UX) |
d23c3b6f | 1704 | Does not automatically flush output handles on some platforms. |
e41182b5 GS |
1705 | |
1706 | =item getlogin | |
1707 | ||
d23c3b6f LM |
1708 | (S<RISC OS>) |
1709 | Not implemented. | |
e41182b5 | 1710 | |
47cd99a4 | 1711 | =item getpgrp |
e41182b5 | 1712 | |
d23c3b6f LM |
1713 | (Win32, VMS, S<RISC OS>) |
1714 | Not implemented. | |
e41182b5 GS |
1715 | |
1716 | =item getppid | |
1717 | ||
d23c3b6f LM |
1718 | (Win32, S<RISC OS>) |
1719 | Not implemented. | |
e41182b5 | 1720 | |
47cd99a4 | 1721 | =item getpriority |
e41182b5 | 1722 | |
d23c3b6f LM |
1723 | (Win32, VMS, S<RISC OS>, VOS) |
1724 | Not implemented. | |
e41182b5 | 1725 | |
47cd99a4 | 1726 | =item getpwnam |
e41182b5 | 1727 | |
d23c3b6f LM |
1728 | (Win32) |
1729 | Not implemented. | |
e41182b5 | 1730 | |
d23c3b6f LM |
1731 | (S<RISC OS>) |
1732 | Not useful. | |
b8099c3d | 1733 | |
47cd99a4 | 1734 | =item getgrnam |
e41182b5 | 1735 | |
d23c3b6f LM |
1736 | (Win32, VMS, S<RISC OS>) |
1737 | Not implemented. | |
e41182b5 | 1738 | |
47cd99a4 | 1739 | =item getnetbyname |
e41182b5 | 1740 | |
d23c3b6f LM |
1741 | (Android, Win32, S<Plan 9>) |
1742 | Not implemented. | |
e41182b5 | 1743 | |
47cd99a4 | 1744 | =item getpwuid |
e41182b5 | 1745 | |
d23c3b6f LM |
1746 | (Win32) |
1747 | Not implemented. | |
e41182b5 | 1748 | |
d23c3b6f LM |
1749 | (S<RISC OS>) |
1750 | Not useful. | |
b8099c3d | 1751 | |
47cd99a4 | 1752 | =item getgrgid |
e41182b5 | 1753 | |
d23c3b6f LM |
1754 | (Win32, VMS, S<RISC OS>) |
1755 | Not implemented. | |
e41182b5 | 1756 | |
47cd99a4 | 1757 | =item getnetbyaddr |
e41182b5 | 1758 | |
d23c3b6f LM |
1759 | (Android, Win32, S<Plan 9>) |
1760 | Not implemented. | |
e41182b5 | 1761 | |
47cd99a4 | 1762 | =item getprotobynumber |
e41182b5 | 1763 | |
d23c3b6f LM |
1764 | (Android) |
1765 | Not implemented. | |
43b08d74 | 1766 | |
e41182b5 GS |
1767 | =item getpwent |
1768 | ||
d23c3b6f LM |
1769 | (Android, Win32) |
1770 | Not implemented. | |
e41182b5 GS |
1771 | |
1772 | =item getgrent | |
1773 | ||
d23c3b6f LM |
1774 | (Android, Win32, VMS) |
1775 | Not implemented. | |
e41182b5 | 1776 | |
ef5a6dd7 JH |
1777 | =item gethostbyname |
1778 | ||
d23c3b6f | 1779 | (S<Irix 5>) |
ef5a6dd7 | 1780 | C<gethostbyname('localhost')> does not work everywhere: you may have |
d23c3b6f | 1781 | to use C<gethostbyname('127.0.0.1')>. |
ef5a6dd7 | 1782 | |
e41182b5 GS |
1783 | =item gethostent |
1784 | ||
d23c3b6f LM |
1785 | (Win32) |
1786 | Not implemented. | |
e41182b5 GS |
1787 | |
1788 | =item getnetent | |
1789 | ||
d23c3b6f LM |
1790 | (Android, Win32, S<Plan 9>) |
1791 | Not implemented. | |
e41182b5 GS |
1792 | |
1793 | =item getprotoent | |
1794 | ||
d23c3b6f LM |
1795 | (Android, Win32, S<Plan 9>) |
1796 | Not implemented. | |
e41182b5 GS |
1797 | |
1798 | =item getservent | |
1799 | ||
d23c3b6f LM |
1800 | (Win32, S<Plan 9>) |
1801 | Not implemented. | |
e41182b5 | 1802 | |
43b08d74 BF |
1803 | =item seekdir |
1804 | ||
d23c3b6f LM |
1805 | (Android) |
1806 | Not implemented. | |
43b08d74 | 1807 | |
47cd99a4 | 1808 | =item sethostent |
e41182b5 | 1809 | |
d23c3b6f LM |
1810 | (Android, Win32, S<Plan 9>, S<RISC OS>) |
1811 | Not implemented. | |
e41182b5 | 1812 | |
47cd99a4 | 1813 | =item setnetent |
e41182b5 | 1814 | |
d23c3b6f LM |
1815 | (Win32, S<Plan 9>, S<RISC OS>) |
1816 | Not implemented. | |
e41182b5 | 1817 | |
47cd99a4 | 1818 | =item setprotoent |
e41182b5 | 1819 | |
d23c3b6f LM |
1820 | (Android, Win32, S<Plan 9>, S<RISC OS>) |
1821 | Not implemented. | |
e41182b5 | 1822 | |
47cd99a4 | 1823 | =item setservent |
e41182b5 | 1824 | |
d23c3b6f LM |
1825 | (S<Plan 9>, Win32, S<RISC OS>) |
1826 | Not implemented. | |
e41182b5 GS |
1827 | |
1828 | =item endpwent | |
1829 | ||
d23c3b6f LM |
1830 | (Win32) |
1831 | Not implemented. | |
e41182b5 | 1832 | |
d23c3b6f LM |
1833 | (Android) |
1834 | Either not implemented or a no-op. | |
43b08d74 | 1835 | |
e41182b5 GS |
1836 | =item endgrent |
1837 | ||
d23c3b6f LM |
1838 | (Android, S<RISC OS>, VMS, Win32) |
1839 | Not implemented. | |
e41182b5 GS |
1840 | |
1841 | =item endhostent | |
1842 | ||
d23c3b6f LM |
1843 | (Android, Win32) |
1844 | Not implemented. | |
e41182b5 GS |
1845 | |
1846 | =item endnetent | |
1847 | ||
d23c3b6f LM |
1848 | (Android, Win32, S<Plan 9>) |
1849 | Not implemented. | |
e41182b5 GS |
1850 | |
1851 | =item endprotoent | |
1852 | ||
d23c3b6f LM |
1853 | (Android, Win32, S<Plan 9>) |
1854 | Not implemented. | |
e41182b5 GS |
1855 | |
1856 | =item endservent | |
1857 | ||
d23c3b6f LM |
1858 | (S<Plan 9>, Win32) |
1859 | Not implemented. | |
e41182b5 | 1860 | |
1096c54b | 1861 | =item getsockopt |
e41182b5 | 1862 | |
d23c3b6f LM |
1863 | (S<Plan 9>) |
1864 | Not implemented. | |
e41182b5 | 1865 | |
e41182b5 GS |
1866 | =item glob |
1867 | ||
83a46a63 LM |
1868 | This operator is implemented via the L<C<File::Glob>|File::Glob> extension |
1869 | on most platforms. See L<File::Glob> for portability information. | |
b8099c3d | 1870 | |
62aa5637 MS |
1871 | =item gmtime |
1872 | ||
83a46a63 LM |
1873 | In theory, C<gmtime> is reliable from -2**63 to 2**63-1. However, |
1874 | because work-arounds in the implementation use floating point numbers, | |
461d5a49 MS |
1875 | it will become inaccurate as the time gets larger. This is a bug and |
1876 | will be fixed in the future. | |
62aa5637 | 1877 | |
d23c3b6f LM |
1878 | (VOS) |
1879 | Time values are 32-bit quantities. | |
10fb90aa | 1880 | |
1096c54b | 1881 | =item ioctl |
e41182b5 | 1882 | |
d23c3b6f LM |
1883 | (VMS) |
1884 | Not implemented. | |
e41182b5 | 1885 | |
d23c3b6f | 1886 | (Win32) |
2c044526 | 1887 | Available only for socket handles, and it does what the C<ioctlsocket()> call |
d23c3b6f | 1888 | in the Winsock API does. |
e41182b5 | 1889 | |
d23c3b6f LM |
1890 | (S<RISC OS>) |
1891 | Available only for socket handles. | |
b8099c3d | 1892 | |
47cd99a4 | 1893 | =item kill |
e41182b5 | 1894 | |
d23c3b6f LM |
1895 | (S<RISC OS>) |
1896 | Not implemented, hence not useful for taint checking. | |
e41182b5 | 1897 | |
d23c3b6f | 1898 | (Win32) |
83a46a63 LM |
1899 | C<kill> doesn't send a signal to the identified process like it does on |
1900 | Unix platforms. Instead C<kill($sig, $pid)> terminates the process | |
1901 | identified by C<$pid>, and makes it exit immediately with exit status | |
1902 | C<$sig>. As in Unix, if C<$sig> is 0 and the specified process exists, it | |
d23c3b6f | 1903 | returns true without actually terminating it. |
e41182b5 | 1904 | |
d23c3b6f | 1905 | (Win32) |
2c044526 | 1906 | C<kill(-9, $pid)> will terminate the process specified by C<$pid> and |
d0302514 JD |
1907 | recursively all child processes owned by it. This is different from |
1908 | the Unix semantics, where the signal will be delivered to all | |
1909 | processes in the same process group as the process specified by | |
d23c3b6f | 1910 | C<$pid>. |
d0302514 | 1911 | |
d23c3b6f | 1912 | (VMS) |
96f902ff | 1913 | A pid of -1 indicating all processes on the system is not currently |
d23c3b6f | 1914 | supported. |
016930a6 | 1915 | |
47cd99a4 | 1916 | =item link |
e41182b5 | 1917 | |
d23c3b6f LM |
1918 | (S<RISC OS>, VOS) |
1919 | Not implemented. | |
e41182b5 | 1920 | |
d23c3b6f | 1921 | (AmigaOS) |
433acd8a | 1922 | Link count not updated because hard links are not quite that hard |
d23c3b6f | 1923 | (They are sort of half-way between hard and soft links). |
433acd8a | 1924 | |
d23c3b6f | 1925 | (Win32) |
63d6c08b JD |
1926 | Hard links are implemented on Win32 under NTFS only. They are |
1927 | natively supported on Windows 2000 and later. On Windows NT they | |
1928 | are implemented using the Windows POSIX subsystem support and the | |
1929 | Perl process will need Administrator or Backup Operator privileges | |
d23c3b6f | 1930 | to create hard links. |
a3dfe201 | 1931 | |
d23c3b6f LM |
1932 | (VMS) |
1933 | Available on 64 bit OpenVMS 8.2 and later. | |
016930a6 | 1934 | |
62aa5637 MS |
1935 | =item localtime |
1936 | ||
83a46a63 LM |
1937 | C<localtime> has the same range as L</gmtime>, but because time zone |
1938 | rules change, its accuracy for historical and future times may degrade | |
dc164757 | 1939 | but usually by no more than an hour. |
62aa5637 | 1940 | |
e41182b5 GS |
1941 | =item lstat |
1942 | ||
d23c3b6f LM |
1943 | (S<RISC OS>) |
1944 | Not implemented. | |
e41182b5 | 1945 | |
d23c3b6f LM |
1946 | (Win32) |
1947 | Return values (especially for device and inode) may be bogus. | |
e41182b5 | 1948 | |
47cd99a4 | 1949 | =item msgctl |
e41182b5 | 1950 | |
47cd99a4 | 1951 | =item msgget |
e41182b5 | 1952 | |
47cd99a4 | 1953 | =item msgsnd |
e41182b5 | 1954 | |
47cd99a4 | 1955 | =item msgrcv |
e41182b5 | 1956 | |
d23c3b6f LM |
1957 | (Android, Win32, VMS, S<Plan 9>, S<RISC OS>, VOS) |
1958 | Not implemented. | |
e41182b5 | 1959 | |
47cd99a4 | 1960 | =item open |
e41182b5 | 1961 | |
d23c3b6f LM |
1962 | (Win32, S<RISC OS>) |
1963 | Open modes C<|-> and C<-|> are unsupported. | |
e41182b5 | 1964 | |
d23c3b6f | 1965 | (SunOS, Solaris, HP-UX) |
0f897271 | 1966 | Opening a process does not automatically flush output handles on some |
d23c3b6f | 1967 | platforms. |
0f897271 | 1968 | |
e41182b5 GS |
1969 | =item readlink |
1970 | ||
d23c3b6f LM |
1971 | (Win32, VMS, S<RISC OS>) |
1972 | Not implemented. | |
e41182b5 | 1973 | |
47cd99a4 | 1974 | =item rename |
c9b2b9d4 | 1975 | |
d23c3b6f LM |
1976 | (Win32) |
1977 | Can't move directories between directories on different logical volumes. | |
c9b2b9d4 | 1978 | |
3ba4b5c1 JD |
1979 | =item rewinddir |
1980 | ||
d23c3b6f | 1981 | (Win32) |
83a46a63 LM |
1982 | Will not cause L<C<readdir>|perlfunc/readdir DIRHANDLE> to re-read the |
1983 | directory stream. The entries already read before the C<rewinddir> call | |
d23c3b6f | 1984 | will just be returned again from a cache buffer. |
3ba4b5c1 | 1985 | |
47cd99a4 | 1986 | =item select |
e41182b5 | 1987 | |
d23c3b6f LM |
1988 | (Win32, VMS) |
1989 | Only implemented on sockets. | |
e41182b5 | 1990 | |
d23c3b6f LM |
1991 | (S<RISC OS>) |
1992 | Only reliable on sockets. | |
b8099c3d | 1993 | |
83a46a63 LM |
1994 | Note that the L<C<select FILEHANDLE>|perlfunc/select FILEHANDLE> form is |
1995 | generally portable. | |
63f87e49 | 1996 | |
47cd99a4 | 1997 | =item semctl |
e41182b5 | 1998 | |
47cd99a4 | 1999 | =item semget |
e41182b5 | 2000 | |
47cd99a4 | 2001 | =item semop |
e41182b5 | 2002 | |
d23c3b6f LM |
2003 | (Android, Win32, VMS, S<RISC OS>) |
2004 | Not implemented. | |
e41182b5 | 2005 | |
a3dfe201 GS |
2006 | =item setgrent |
2007 | ||
d23c3b6f LM |
2008 | (Android, VMS, Win32, S<RISC OS>) |
2009 | Not implemented. | |
a3dfe201 | 2010 | |
47cd99a4 | 2011 | =item setpgrp |
e41182b5 | 2012 | |
d23c3b6f LM |
2013 | (Win32, VMS, S<RISC OS>, VOS) |
2014 | Not implemented. | |
e41182b5 | 2015 | |
47cd99a4 | 2016 | =item setpriority |
e41182b5 | 2017 | |
d23c3b6f LM |
2018 | (Win32, VMS, S<RISC OS>, VOS) |
2019 | Not implemented. | |
e41182b5 | 2020 | |
a3dfe201 GS |
2021 | =item setpwent |
2022 | ||
d23c3b6f LM |
2023 | (Android, Win32, S<RISC OS>) |
2024 | Not implemented. | |
a3dfe201 | 2025 | |
47cd99a4 | 2026 | =item setsockopt |
e41182b5 | 2027 | |
d23c3b6f LM |
2028 | (S<Plan 9>) |
2029 | Not implemented. | |
e41182b5 | 2030 | |
47cd99a4 | 2031 | =item shmctl |
e41182b5 | 2032 | |
47cd99a4 | 2033 | =item shmget |
e41182b5 | 2034 | |
47cd99a4 | 2035 | =item shmread |
e41182b5 | 2036 | |
47cd99a4 | 2037 | =item shmwrite |
e41182b5 | 2038 | |
d23c3b6f LM |
2039 | (Android, Win32, VMS, S<RISC OS>) |
2040 | Not implemented. | |
e41182b5 | 2041 | |
001e9f89 DD |
2042 | =item sleep |
2043 | ||
d23c3b6f | 2044 | (Win32) |
3cd50447 | 2045 | Emulated using synchronization functions such that it can be |
83a46a63 | 2046 | interrupted by L<C<alarm>|perlfunc/alarm SECONDS>, and limited to a |
d23c3b6f | 2047 | maximum of 4294967 seconds, approximately 49 days. |
80cbd5ad | 2048 | |
47cd99a4 | 2049 | =item socketpair |
e41182b5 | 2050 | |
d23c3b6f LM |
2051 | (S<RISC OS>) |
2052 | Not implemented. | |
10fb90aa | 2053 | |
d23c3b6f LM |
2054 | (VMS) |
2055 | Available on 64 bit OpenVMS 8.2 and later. | |
e41182b5 | 2056 | |
e41182b5 GS |
2057 | =item stat |
2058 | ||
83a46a63 LM |
2059 | Platforms that do not have C<rdev>, C<blksize>, or C<blocks> will return |
2060 | these as C<''>, so numeric comparison or manipulation of these fields may | |
2061 | cause 'not numeric' warnings. | |
d62e1b7f | 2062 | |
d23c3b6f LM |
2063 | (S<Mac OS X>) |
2064 | C<ctime> not supported on UFS. | |
e41182b5 | 2065 | |
d23c3b6f LM |
2066 | (Win32) |
2067 | C<ctime> is creation time instead of inode change time. | |
95a3fe12 | 2068 | |
d23c3b6f LM |
2069 | (Win32) |
2070 | C<dev> and C<ino> are not meaningful. | |
e41182b5 | 2071 | |
d23c3b6f LM |
2072 | (VMS) |
2073 | C<dev> and C<ino> are not necessarily reliable. | |
e41182b5 | 2074 | |
d23c3b6f | 2075 | (S<RISC OS>) |
83a46a63 | 2076 | C<mtime>, C<atime> and C<ctime> all return the last modification time. |
d23c3b6f | 2077 | C<dev> and C<ino> are not necessarily reliable. |
b8099c3d | 2078 | |
d23c3b6f | 2079 | (OS/2) |
83a46a63 | 2080 | C<dev>, C<rdev>, C<blksize>, and C<blocks> are not available. C<ino> is not |
d23c3b6f | 2081 | meaningful and will differ between stat calls on the same file. |
d62e1b7f | 2082 | |
d23c3b6f | 2083 | (Cygwin) |
83a46a63 | 2084 | Some versions of cygwin when doing a C<stat("foo")> and not finding it |
d23c3b6f | 2085 | may then attempt to C<stat("foo.exe")>. |
73e9292c | 2086 | |
d23c3b6f | 2087 | (Win32) |
83a46a63 | 2088 | C<stat> needs to open the file to determine the link count |
1fafdf34 | 2089 | and update attributes that may have been changed through hard links. |
83a46a63 | 2090 | Setting L<C<${^WIN32_SLOPPY_STAT}>|perlvar/${^WIN32_SLOPPY_STAT}> to a |
d23c3b6f | 2091 | true value speeds up C<stat> by not performing this operation. |
1fafdf34 | 2092 | |
47cd99a4 | 2093 | =item symlink |
e41182b5 | 2094 | |
d23c3b6f LM |
2095 | (Win32, S<RISC OS>) |
2096 | Not implemented. | |
c73b03b7 | 2097 | |
d23c3b6f | 2098 | (VMS) |
c73b03b7 | 2099 | Implemented on 64 bit VMS 8.3. VMS requires the symbolic link to be in Unix |
d23c3b6f | 2100 | syntax if it is intended to resolve to a valid path. |
e41182b5 | 2101 | |
47cd99a4 | 2102 | =item syscall |
e41182b5 | 2103 | |
d23c3b6f LM |
2104 | (Win32, VMS, S<RISC OS>, VOS) |
2105 | Not implemented. | |
e41182b5 | 2106 | |
47cd99a4 | 2107 | =item sysopen |
f34d0673 | 2108 | |
d23c3b6f | 2109 | (S<Mac OS>, OS/390) |
83a46a63 LM |
2110 | The traditional C<0>, C<1>, and C<2> MODEs are implemented with different |
2111 | numeric values on some systems. The flags exported by L<C<Fcntl>|Fcntl> | |
2112 | (C<O_RDONLY>, C<O_WRONLY>, C<O_RDWR>) should work everywhere though. | |
f34d0673 | 2113 | |
47cd99a4 | 2114 | =item system |
e41182b5 | 2115 | |
d23c3b6f | 2116 | (Win32) |
e41182b5 | 2117 | As an optimization, may not call the command shell specified in |
b7df3edc | 2118 | C<$ENV{PERL5SHELL}>. C<system(1, @args)> spawns an external |
e41182b5 GS |
2119 | process and immediately returns its process designator, without |
2120 | waiting for it to terminate. Return value may be used subsequently | |
83a46a63 LM |
2121 | in L<C<wait>|perlfunc/wait> or L<C<waitpid>|perlfunc/waitpid PID,FLAGS>. |
2122 | Failure to C<spawn()> a subprocess is indicated by setting | |
2123 | L<C<$?>|perlvar/$?> to C<<< 255 << 8 >>>. L<C<$?>|perlvar/$?> is set in a | |
2124 | way compatible with Unix (i.e. the exit status of the subprocess is | |
d23c3b6f | 2125 | obtained by C<<< $? >> 8 >>>, as described in the documentation). |
e41182b5 | 2126 | |
d23c3b6f | 2127 | (S<RISC OS>) |
b8099c3d CN |
2128 | There is no shell to process metacharacters, and the native standard is |
2129 | to pass a command line terminated by "\n" "\r" or "\0" to the spawned | |
c47ff5f1 | 2130 | program. Redirection such as C<< > foo >> is performed (if at all) by |
83a46a63 LM |
2131 | the run time library of the spawned program. C<system LIST> will call |
2132 | the Unix emulation library's L<C<exec>|perlfunc/exec LIST> emulation, | |
2133 | which attempts to provide emulation of the stdin, stdout, stderr in force | |
2134 | in the parent, provided the child program uses a compatible version of the | |
2135 | emulation library. C<system SCALAR> will call the native command line | |
2136 | directly and no such emulation of a child Unix program will occur. | |
d23c3b6f | 2137 | Mileage B<will> vary. |
b8099c3d | 2138 | |
d23c3b6f | 2139 | (Win32) |
94d4006a | 2140 | C<system LIST> without the use of indirect object syntax (C<system PROGRAM LIST>) |
d23c3b6f | 2141 | may fall back to trying the shell if the first C<spawn()> fails. |
94d4006a | 2142 | |
0f897271 | 2143 | (SunOS, Solaris, HP-UX) |
d23c3b6f | 2144 | Does not automatically flush output handles on some platforms. |
0f897271 | 2145 | |
d23c3b6f | 2146 | (VMS) |
b3b7afb8 CB |
2147 | As with Win32, C<system(1, @args)> spawns an external process and |
2148 | immediately returns its process designator without waiting for the | |
2149 | process to terminate. In this case the return value may be used subsequently | |
2150 | in L<C<wait>|perlfunc/wait> or L<C<waitpid>|perlfunc/waitpid PID,FLAGS>. | |
2151 | Otherwise the return value is POSIX-like (shifted up by 8 bits), which only | |
2152 | allows room for a made-up value derived from the severity bits of the native | |
83a46a63 LM |
2153 | 32-bit condition code (unless overridden by |
2154 | L<C<use vmsish 'status'>|vmsish/C<vmsish status>>). If the native | |
2155 | condition code is one that has a POSIX value encoded, the POSIX value will | |
2156 | be decoded to extract the expected exit value. For more details see | |
d23c3b6f | 2157 | L<perlvms/$?>. |
9bc98430 | 2158 | |
43b08d74 BF |
2159 | =item telldir |
2160 | ||
d23c3b6f LM |
2161 | (Android) |
2162 | Not implemented. | |
43b08d74 | 2163 | |
e41182b5 GS |
2164 | =item times |
2165 | ||
d23c3b6f | 2166 | (Win32) |
83a46a63 | 2167 | "Cumulative" times will be bogus. On anything other than Windows NT |
63f87e49 | 2168 | or Windows 2000, "system" time will be bogus, and "user" time is |
83a46a63 | 2169 | actually the time returned by the L<C<clock()>|clock(3)> function in the C |
d23c3b6f | 2170 | runtime library. |
e41182b5 | 2171 | |
d23c3b6f LM |
2172 | (S<RISC OS>) |
2173 | Not useful. | |
b8099c3d | 2174 | |
47cd99a4 | 2175 | =item truncate |
e41182b5 | 2176 | |
d23c3b6f LM |
2177 | (Older versions of VMS) |
2178 | Not implemented. | |
e41182b5 | 2179 | |
d23c3b6f LM |
2180 | (VOS) |
2181 | Truncation to same-or-shorter lengths only. | |
495c5fdc | 2182 | |
d23c3b6f | 2183 | (Win32) |
4cfdb94f | 2184 | If a FILEHANDLE is supplied, it must be writable and opened in append |
83a46a63 LM |
2185 | mode (i.e., use C<<< open(my $fh, '>>', 'filename') >>> |
2186 | or C<sysopen(my $fh, ..., O_APPEND|O_RDWR)>. If a filename is supplied, it | |
d23c3b6f | 2187 | should not be held open elsewhere. |
4cfdb94f | 2188 | |
e41182b5 GS |
2189 | =item umask |
2190 | ||
83a46a63 | 2191 | Returns C<undef> where unavailable. |
e41182b5 | 2192 | |
d23c3b6f | 2193 | (AmigaOS) |
b7df3edc | 2194 | C<umask> works but the correct permissions are set only when the file |
d23c3b6f | 2195 | is finally closed. |
433acd8a | 2196 | |
47cd99a4 | 2197 | =item utime |
e41182b5 | 2198 | |
d23c3b6f LM |
2199 | (VMS, S<RISC OS>) |
2200 | Only the modification time is updated. | |
e41182b5 | 2201 | |
d23c3b6f | 2202 | (Win32) |
322422de | 2203 | May not behave as expected. Behavior depends on the C runtime |
83a46a63 LM |
2204 | library's implementation of L<C<utime()>|utime(2)>, and the filesystem |
2205 | being used. The FAT filesystem typically does not support an "access | |
2206 | time" field, and it may limit timestamps to a granularity of two seconds. | |
e41182b5 GS |
2207 | |
2208 | =item wait | |
2209 | ||
47cd99a4 | 2210 | =item waitpid |
e41182b5 | 2211 | |
d23c3b6f | 2212 | (Win32) |
e41182b5 | 2213 | Can only be applied to process handles returned for processes spawned |
83a46a63 | 2214 | using C<system(1, ...)> or pseudo processes created with |
d23c3b6f | 2215 | L<C<fork>|perlfunc/fork>. |
e41182b5 | 2216 | |
d23c3b6f LM |
2217 | (S<RISC OS>) |
2218 | Not useful. | |
b8099c3d | 2219 | |
e41182b5 GS |
2220 | =back |
2221 | ||
2222 | ||
7c35b6af | 2223 | =head1 Supported Platforms |
ba58ab26 | 2224 | |
7c35b6af RGS |
2225 | The following platforms are known to build Perl 5.12 (as of April 2010, |
2226 | its release date) from the standard source code distribution available | |
500f1b69 | 2227 | at L<http://www.cpan.org/src> |
bb377ba2 | 2228 | |
bb377ba2 JV |
2229 | =over |
2230 | ||
2231 | =item Linux (x86, ARM, IA64) | |
2232 | ||
e0d9a2c8 | 2233 | =item HP-UX |
bb377ba2 JV |
2234 | |
2235 | =item AIX | |
2236 | ||
2237 | =item Win32 | |
2238 | ||
2239 | =over | |
2240 | ||
2241 | =item Windows 2000 | |
2242 | ||
2243 | =item Windows XP | |
2244 | ||
2245 | =item Windows Server 2003 | |
2246 | ||
2247 | =item Windows Vista | |
2248 | ||
2249 | =item Windows Server 2008 | |
2250 | ||
3b665c47 JD |
2251 | =item Windows 7 |
2252 | ||
bb377ba2 JV |
2253 | =back |
2254 | ||
2d9ede6e JH |
2255 | =item Cygwin |
2256 | ||
67e52905 TC |
2257 | Some tests are known to fail: |
2258 | ||
2259 | =over | |
2260 | ||
2261 | =item * | |
2262 | ||
cb0ee57a | 2263 | F<ext/XS-APItest/t/call_checker.t> - see |
67e52905 TC |
2264 | L<https://rt.perl.org/Ticket/Display.html?id=78502> |
2265 | ||
2266 | =item * | |
2267 | ||
2268 | F<dist/I18N-Collate/t/I18N-Collate.t> | |
2269 | ||
2270 | =item * | |
2271 | ||
2272 | F<ext/Win32CORE/t/win32core.t> - may fail on recent cygwin installs. | |
2273 | ||
2274 | =back | |
2275 | ||
bb377ba2 JV |
2276 | =item Solaris (x86, SPARC) |
2277 | ||
1b0ab010 JV |
2278 | =item OpenVMS |
2279 | ||
2280 | =over | |
2281 | ||
2282 | =item Alpha (7.2 and later) | |
2283 | ||
2284 | =item I64 (8.2 and later) | |
2285 | ||
2286 | =back | |
bb377ba2 JV |
2287 | |
2288 | =item Symbian | |
2289 | ||
2290 | =item NetBSD | |
2291 | ||
2292 | =item FreeBSD | |
2293 | ||
2d8e9a35 CBW |
2294 | =item Debian GNU/kFreeBSD |
2295 | ||
bb377ba2 JV |
2296 | =item Haiku |
2297 | ||
2298 | =item Irix (6.5. What else?) | |
2299 | ||
2300 | =item OpenBSD | |
2301 | ||
2302 | =item Dragonfly BSD | |
2303 | ||
b60fc215 CBW |
2304 | =item Midnight BSD |
2305 | ||
a62bfce3 CBW |
2306 | =item QNX Neutrino RTOS (6.5.0) |
2307 | ||
bb377ba2 JV |
2308 | =item MirOS BSD |
2309 | ||
7d4dfb6d PG |
2310 | =item Stratus OpenVOS (17.0 or later) |
2311 | ||
bb377ba2 JV |
2312 | Caveats: |
2313 | ||
2314 | =over | |
2315 | ||
2316 | =item time_t issues that may or may not be fixed | |
2317 | ||
2318 | =back | |
2319 | ||
bb377ba2 JV |
2320 | =item Symbian (Series 60 v3, 3.2 and 5 - what else?) |
2321 | ||
10fb90aa | 2322 | =item Stratus VOS / OpenVOS |
bb377ba2 JV |
2323 | |
2324 | =item AIX | |
2325 | ||
25be0a81 BF |
2326 | =item Android |
2327 | ||
dcfa7505 TC |
2328 | =item FreeMINT |
2329 | ||
2330 | Perl now builds with FreeMiNT/Atari. It fails a few tests, that needs | |
2331 | some investigation. | |
2332 | ||
2333 | The FreeMiNT port uses GNU dld for loadable module capabilities. So | |
2334 | ensure you have that library installed when building perl. | |
2335 | ||
bb377ba2 JV |
2336 | =back |
2337 | ||
2890cc8c | 2338 | =head1 EOL Platforms |
7b0e9f13 AD |
2339 | |
2340 | =head2 (Perl 5.20) | |
bb377ba2 JV |
2341 | |
2342 | The following platforms were supported by a previous version of | |
2343 | Perl but have been officially removed from Perl's source code | |
7b0e9f13 | 2344 | as of 5.20: |
bb377ba2 JV |
2345 | |
2346 | =over | |
2347 | ||
7b0e9f13 | 2348 | =item AT&T 3b1 |
bb377ba2 JV |
2349 | |
2350 | =back | |
2351 | ||
7b0e9f13 AD |
2352 | =head2 (Perl 5.14) |
2353 | ||
8cbe99e5 JD |
2354 | The following platforms were supported up to 5.10. They may still |
2355 | have worked in 5.12, but supporting code has been removed for 5.14: | |
bb377ba2 JV |
2356 | |
2357 | =over | |
2358 | ||
2359 | =item Windows 95 | |
2360 | ||
2361 | =item Windows 98 | |
2362 | ||
2363 | =item Windows ME | |
2364 | ||
2365 | =item Windows NT4 | |
2366 | ||
2367 | =back | |
2368 | ||
7b0e9f13 AD |
2369 | =head2 (Perl 5.12) |
2370 | ||
2371 | The following platforms were supported by a previous version of | |
2372 | Perl but have been officially removed from Perl's source code | |
2373 | as of 5.12: | |
2374 | ||
2375 | =over | |
2376 | ||
2377 | =item Atari MiNT | |
2378 | ||
2379 | =item Apollo Domain/OS | |
2380 | ||
2381 | =item Apple Mac OS 8/9 | |
2382 | ||
2383 | =item Tenon Machten | |
2384 | ||
2385 | =back | |
2386 | ||
2387 | ||
bb377ba2 JV |
2388 | =head1 Supported Platforms (Perl 5.8) |
2389 | ||
2390 | As of July 2002 (the Perl release 5.8.0), the following platforms were | |
cec2c193 | 2391 | able to build Perl from the standard source code distribution |
500f1b69 | 2392 | available at L<http://www.cpan.org/src/> |
cec2c193 JH |
2393 | |
2394 | AIX | |
2395 | BeOS | |
6f683aa2 | 2396 | BSD/OS (BSDi) |
cec2c193 | 2397 | Cygwin |
ea297d26 | 2398 | DG/UX |
811b48f2 | 2399 | DOS DJGPP 1) |
cec2c193 JH |
2400 | DYNIX/ptx |
2401 | EPOC R5 | |
2402 | FreeBSD | |
6f683aa2 | 2403 | HI-UXMPP (Hitachi) (5.8.0 worked but we didn't know it) |
cec2c193 JH |
2404 | HP-UX |
2405 | IRIX | |
2406 | Linux | |
8939ba94 | 2407 | Mac OS Classic |
6f683aa2 | 2408 | Mac OS X (Darwin) |
cec2c193 JH |
2409 | MPE/iX |
2410 | NetBSD | |
2411 | NetWare | |
2412 | NonStop-UX | |
6f683aa2 | 2413 | ReliantUNIX (formerly SINIX) |
cec2c193 | 2414 | OpenBSD |
6f683aa2 | 2415 | OpenVMS (formerly VMS) |
3ebac25b | 2416 | Open UNIX (Unixware) (since Perl 5.8.1/5.9.0) |
cec2c193 | 2417 | OS/2 |
522b859a | 2418 | OS/400 (using the PASE) (since Perl 5.8.1/5.9.0) |
6f683aa2 | 2419 | POSIX-BC (formerly BS2000) |
cec2c193 JH |
2420 | QNX |
2421 | Solaris | |
70de81db | 2422 | SunOS 4 |
6f683aa2 JH |
2423 | SUPER-UX (NEC) |
2424 | Tru64 UNIX (formerly DEC OSF/1, Digital UNIX) | |
cec2c193 JH |
2425 | UNICOS |
2426 | UNICOS/mk | |
2427 | UTS | |
7d4dfb6d | 2428 | VOS / OpenVOS |
811b48f2 | 2429 | Win95/98/ME/2K/XP 2) |
c40b5d1d | 2430 | WinCE |
6f683aa2 | 2431 | z/OS (formerly OS/390) |
cec2c193 | 2432 | VM/ESA |
ba58ab26 | 2433 | |
811b48f2 JH |
2434 | 1) in DOS mode either the DOS or OS/2 ports can be used |
2435 | 2) compilers: Borland, MinGW (GCC), VC6 | |
cec2c193 | 2436 | |
c40b5d1d | 2437 | The following platforms worked with the previous releases (5.6 and |
cec2c193 JH |
2438 | 5.7), but we did not manage either to fix or to test these in time |
2439 | for the 5.8.0 release. There is a very good chance that many of these | |
70de81db | 2440 | will work fine with the 5.8.0. |
cec2c193 | 2441 | |
8da2b1be | 2442 | BSD/OS |
cec2c193 JH |
2443 | DomainOS |
2444 | Hurd | |
2445 | LynxOS | |
2446 | MachTen | |
2447 | PowerMAX | |
2448 | SCO SV | |
cec2c193 JH |
2449 | SVR4 |
2450 | Unixware | |
2451 | Windows 3.1 | |
ba58ab26 | 2452 | |
70de81db JH |
2453 | Known to be broken for 5.8.0 (but 5.6.1 and 5.7.2 can be used): |
2454 | ||
61988e87 | 2455 | AmigaOS 3 |
70de81db | 2456 | |
ba58ab26 | 2457 | The following platforms have been known to build Perl from source in |
fd46a41b JH |
2458 | the past (5.005_03 and earlier), but we haven't been able to verify |
2459 | their status for the current release, either because the | |
2460 | hardware/software platforms are rare or because we don't have an | |
2461 | active champion on these platforms--or both. They used to work, | |
2462 | though, so go ahead and try compiling them, and let perlbug@perl.org | |
2463 | of any trouble. | |
ba58ab26 | 2464 | |
cec2c193 JH |
2465 | 3b1 |
2466 | A/UX | |
cec2c193 JH |
2467 | ConvexOS |
2468 | CX/UX | |
2469 | DC/OSx | |
2470 | DDE SMES | |
2471 | DOS EMX | |
2472 | Dynix | |
2473 | EP/IX | |
2474 | ESIX | |
2475 | FPS | |
2476 | GENIX | |
2477 | Greenhills | |
2478 | ISC | |
2479 | MachTen 68k | |
cec2c193 JH |
2480 | MPC |
2481 | NEWS-OS | |
2482 | NextSTEP | |
2483 | OpenSTEP | |
2484 | Opus | |
2485 | Plan 9 | |
cec2c193 | 2486 | RISC/os |
8da2b1be | 2487 | SCO ODT/OSR |
cec2c193 JH |
2488 | Stellar |
2489 | SVR2 | |
2490 | TI1500 | |
2491 | TitanOS | |
2492 | Ultrix | |
2493 | Unisys Dynix | |
ba58ab26 JH |
2494 | |
2495 | The following platforms have their own source code distributions and | |
500f1b69 | 2496 | binaries available via L<http://www.cpan.org/ports/> |
ba58ab26 | 2497 | |
cec2c193 | 2498 | Perl release |
ba58ab26 | 2499 | |
522b859a | 2500 | OS/400 (ILE) 5.005_02 |
cec2c193 | 2501 | Tandem Guardian 5.004 |
ba58ab26 JH |
2502 | |
2503 | The following platforms have only binaries available via | |
500f1b69 | 2504 | L<http://www.cpan.org/ports/index.html> : |
ba58ab26 | 2505 | |
cec2c193 | 2506 | Perl release |
ba58ab26 | 2507 | |
cec2c193 JH |
2508 | Acorn RISCOS 5.005_02 |
2509 | AOS 5.002 | |
2510 | LynxOS 5.004_02 | |
ba58ab26 JH |
2511 | |
2512 | Although we do suggest that you always build your own Perl from | |
2513 | the source code, both for maximal configurability and for security, | |
2514 | in case you are in a hurry you can check | |
500f1b69 | 2515 | L<http://www.cpan.org/ports/index.html> for binary distributions. |
ba58ab26 | 2516 | |
c997b287 GS |
2517 | =head1 SEE ALSO |
2518 | ||
b6c36746 | 2519 | L<perlaix>, L<perlamiga>, L<perlbs2000>, |
8d87852b | 2520 | L<perlce>, L<perlcygwin>, L<perldos>, |
469e7be4 | 2521 | L<perlebcdic>, L<perlfreebsd>, L<perlhurd>, L<perlhpux>, L<perlirix>, |
b5afd346 | 2522 | L<perlmacos>, L<perlmacosx>, |
522b859a JH |
2523 | L<perlnetware>, L<perlos2>, L<perlos390>, L<perlos400>, |
2524 | L<perlplan9>, L<perlqnx>, L<perlsolaris>, L<perltru64>, | |
043fec90 | 2525 | L<perlunicode>, L<perlvms>, L<perlvos>, L<perlwin32>, and L<Win32>. |
c997b287 | 2526 | |
e41182b5 GS |
2527 | =head1 AUTHORS / CONTRIBUTORS |
2528 | ||
de69c9af | 2529 | Abigail <abigail@abigail.be>, |
c47ff5f1 GS |
2530 | Charles Bailey <bailey@newman.upenn.edu>, |
2531 | Graham Barr <gbarr@pobox.com>, | |
2532 | Tom Christiansen <tchrist@perl.com>, | |
06e9666b | 2533 | Nicholas Clark <nick@ccl4.org>, |
c47ff5f1 | 2534 | Thomas Dorner <Thomas.Dorner@start.de>, |
06e9666b A |
2535 | Andy Dougherty <doughera@lafayette.edu>, |
2536 | Dominic Dunlop <domo@computer.org>, | |
2537 | Neale Ferguson <neale@vma.tabnsw.com.au>, | |
c47ff5f1 | 2538 | David J. Fiander <davidf@mks.com>, |
3fd80bd6 | 2539 | Paul Green <Paul.Green@stratus.com>, |
06e9666b | 2540 | M.J.T. Guy <mjtg@cam.ac.uk>, |
61f30a5e | 2541 | Jarkko Hietaniemi <jhi@iki.fi>, |
c47ff5f1 | 2542 | Luther Huffman <lutherh@stratcom.com>, |
06e9666b A |
2543 | Nick Ing-Simmons <nick@ing-simmons.net>, |
2544 | Andreas J. KE<ouml>nig <a.koenig@mind.de>, | |
c47ff5f1 GS |
2545 | Markus Laker <mlaker@contax.co.uk>, |
2546 | Andrew M. Langmead <aml@world.std.com>, | |
83a46a63 | 2547 | Lukas Mai <l.mai@web.de>, |
c47ff5f1 GS |
2548 | Larry Moore <ljmoore@freespace.net>, |
2549 | Paul Moore <Paul.Moore@uk.origin-it.com>, | |
2550 | Chris Nandor <pudge@pobox.com>, | |
1afc07ec | 2551 | Matthias Neeracher <neeracher@mac.com>, |
e71a7dc8 | 2552 | Philip Newton <pne@cpan.org>, |
c47ff5f1 GS |
2553 | Gary Ng <71564.1743@CompuServe.COM>, |
2554 | Tom Phoenix <rootbeer@teleport.com>, | |
2555 | AndrE<eacute> Pirard <A.Pirard@ulg.ac.be>, | |
2556 | Peter Prymmer <pvhp@forte.com>, | |
2557 | Hugo van der Sanden <hv@crypt0.demon.co.uk>, | |
2558 | Gurusamy Sarathy <gsar@activestate.com>, | |
2559 | Paul J. Schinder <schinder@pobox.com>, | |
2560 | Michael G Schwern <schwern@pobox.com>, | |
06e9666b | 2561 | Dan Sugalski <dan@sidhe.org>, |
bbe548ff | 2562 | Nathan Torkington <gnat@frii.com>, |
016930a6 | 2563 | John Malmberg <wb8tyw@qsl.net> |