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 | |
7ee27b7c | 70 | portability (L<"ISSUES">), platform-specific issues (L<"PLATFORMS">), and |
2c044526 | 71 | built-in Perl functions that behave differently on various ports |
7ee27b7c | 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 | |
2c044526 KW |
98 | To trim trailing newlines from text lines use C<chomp()>. With default |
99 | settings that function looks for a trailing C<\n> character and thus | |
31bb23b1 | 100 | trims in a portable way. |
5b3eff12 MS |
101 | |
102 | When dealing with binary files (or text files in binary mode) be sure | |
103 | to explicitly set $/ to the appropriate value for your file format | |
2c044526 | 104 | before using C<chomp()>. |
5b3eff12 | 105 | |
b7df3edc GS |
106 | Because of the "text" mode translation, DOSish perls have limitations |
107 | in using C<seek> and C<tell> on a file accessed in "text" mode. | |
108 | Stick to C<seek>-ing to locations you got from C<tell> (and no | |
109 | others), and you are usually free to use C<seek> and C<tell> even | |
110 | in "text" mode. Using C<seek> or C<tell> or other file operations | |
111 | may be non-portable. If you use C<binmode> on a file, however, you | |
2c044526 | 112 | can usually C<seek> and C<tell> with arbitrary values safely. |
e41182b5 | 113 | |
2c044526 | 114 | A common misconception in socket programming is that S<C<\n eq \012>> |
0a47030a | 115 | everywhere. When using protocols such as common Internet protocols, |
e41182b5 GS |
116 | C<\012> and C<\015> are called for specifically, and the values of |
117 | the logical C<\n> and C<\r> (carriage return) are not reliable. | |
118 | ||
119 | print SOCKET "Hi there, client!\r\n"; # WRONG | |
120 | print SOCKET "Hi there, client!\015\012"; # RIGHT | |
121 | ||
0a47030a GS |
122 | However, using C<\015\012> (or C<\cM\cJ>, or C<\x0D\x0A>) can be tedious |
123 | and unsightly, as well as confusing to those maintaining the code. As | |
2c044526 | 124 | such, the C<Socket> module supplies the Right Thing for those who want it. |
e41182b5 GS |
125 | |
126 | use Socket qw(:DEFAULT :crlf); | |
127 | print SOCKET "Hi there, client!$CRLF" # RIGHT | |
128 | ||
6ab3f9cb | 129 | When reading from a socket, remember that the default input record |
b7df3edc GS |
130 | separator C<$/> is C<\n>, but robust socket code will recognize as |
131 | either C<\012> or C<\015\012> as end of line: | |
e41182b5 | 132 | |
2c044526 | 133 | while (<SOCKET>) { # NOT ADVISABLE! |
e41182b5 GS |
134 | # ... |
135 | } | |
136 | ||
b7df3edc GS |
137 | Because both CRLF and LF end in LF, the input record separator can |
138 | be set to LF and any CR stripped later. Better to write: | |
e41182b5 GS |
139 | |
140 | use Socket qw(:DEFAULT :crlf); | |
141 | local($/) = LF; # not needed if $/ is already \012 | |
142 | ||
143 | while (<SOCKET>) { | |
144 | s/$CR?$LF/\n/; # not sure if socket uses LF or CRLF, OK | |
145 | # s/\015?\012/\n/; # same thing | |
146 | } | |
147 | ||
b7df3edc GS |
148 | This example is preferred over the previous one--even for Unix |
149 | platforms--because now any C<\015>'s (C<\cM>'s) are stripped out | |
e41182b5 GS |
150 | (and there was much rejoicing). |
151 | ||
6ab3f9cb | 152 | Similarly, functions that return text data--such as a function that |
b7df3edc GS |
153 | fetches a web page--should sometimes translate newlines before |
154 | returning the data, if they've not yet been translated to the local | |
155 | newline representation. A single line of code will often suffice: | |
2ee0eb3c | 156 | |
b7df3edc GS |
157 | $data =~ s/\015?\012/\n/g; |
158 | return $data; | |
2ee0eb3c | 159 | |
6ab3f9cb GS |
160 | Some of this may be confusing. Here's a handy reference to the ASCII CR |
161 | and LF characters. You can print it out and stick it in your wallet. | |
162 | ||
74555b7a PP |
163 | LF eq \012 eq \x0A eq \cJ eq chr(10) eq ASCII 10 |
164 | CR eq \015 eq \x0D eq \cM eq chr(13) eq ASCII 13 | |
6ab3f9cb GS |
165 | |
166 | | Unix | DOS | Mac | | |
167 | --------------------------- | |
168 | \n | LF | LF | CR | | |
169 | \r | CR | CR | LF | | |
170 | \n * | LF | CRLF | CR | | |
171 | \r * | CR | CR | LF | | |
172 | --------------------------- | |
173 | * text-mode STDIO | |
174 | ||
b7df3edc GS |
175 | The Unix column assumes that you are not accessing a serial line |
176 | (like a tty) in canonical mode. If you are, then CR on input becomes | |
177 | "\n", and "\n" on output becomes CRLF. | |
178 | ||
6ab3f9cb | 179 | These are just the most common definitions of C<\n> and C<\r> in Perl. |
522b859a JH |
180 | There may well be others. For example, on an EBCDIC implementation |
181 | such as z/OS (OS/390) or OS/400 (using the ILE, the PASE is ASCII-based) | |
182 | the above material is similar to "Unix" but the code numbers change: | |
74555b7a | 183 | |
d770bc45 TS |
184 | LF eq \025 eq \x15 eq \cU eq chr(21) eq CP-1047 21 |
185 | LF eq \045 eq \x25 eq chr(37) eq CP-0037 37 | |
74555b7a PP |
186 | CR eq \015 eq \x0D eq \cM eq chr(13) eq CP-1047 13 |
187 | CR eq \015 eq \x0D eq \cM eq chr(13) eq CP-0037 13 | |
188 | ||
189 | | z/OS | OS/400 | | |
190 | ---------------------- | |
191 | \n | LF | LF | | |
192 | \r | CR | CR | | |
193 | \n * | LF | LF | | |
194 | \r * | CR | CR | | |
195 | ---------------------- | |
196 | * text-mode STDIO | |
6ab3f9cb | 197 | |
322422de GS |
198 | =head2 Numbers endianness and Width |
199 | ||
200 | Different CPUs store integers and floating point numbers in different | |
201 | orders (called I<endianness>) and widths (32-bit and 64-bit being the | |
b7df3edc GS |
202 | most common today). This affects your programs when they attempt to transfer |
203 | numbers in binary format from one CPU architecture to another, | |
204 | usually either "live" via network connection, or by storing the | |
205 | numbers to secondary storage such as a disk file or tape. | |
322422de | 206 | |
2c044526 | 207 | Conflicting storage orders make an utter mess out of the numbers. If a |
d1e3b762 | 208 | little-endian host (Intel, VAX) stores 0x12345678 (305419896 in |
b84d4f81 JH |
209 | decimal), a big-endian host (Motorola, Sparc, PA) reads it as |
210 | 0x78563412 (2018915346 in decimal). Alpha and MIPS can be either: | |
211 | Digital/Compaq used/uses them in little-endian mode; SGI/Cray uses | |
212 | them in big-endian mode. To avoid this problem in network (socket) | |
213 | connections use the C<pack> and C<unpack> formats C<n> and C<N>, the | |
214 | "network" orders. These are guaranteed to be portable. | |
322422de | 215 | |
2c044526 | 216 | As of Perl 5.10.0, you can also use the C<E<gt>> and C<E<lt>> modifiers |
1109a392 MHM |
217 | to force big- or little-endian byte-order. This is useful if you want |
218 | to store signed integers or 64-bit integers, for example. | |
219 | ||
d1e3b762 GS |
220 | You can explore the endianness of your platform by unpacking a |
221 | data structure packed in native format such as: | |
222 | ||
223 | print unpack("h*", pack("s2", 1, 2)), "\n"; | |
224 | # '10002000' on e.g. Intel x86 or Alpha 21064 in little-endian mode | |
225 | # '00100020' on e.g. Motorola 68040 | |
226 | ||
227 | If you need to distinguish between endian architectures you could use | |
228 | either of the variables set like so: | |
229 | ||
230 | $is_big_endian = unpack("h*", pack("s", 1)) =~ /01/; | |
4375e838 | 231 | $is_little_endian = unpack("h*", pack("s", 1)) =~ /^1/; |
d1e3b762 | 232 | |
b7df3edc GS |
233 | Differing widths can cause truncation even between platforms of equal |
234 | endianness. The platform of shorter width loses the upper parts of the | |
322422de GS |
235 | number. There is no good solution for this problem except to avoid |
236 | transferring or storing raw binary numbers. | |
237 | ||
b7df3edc | 238 | One can circumnavigate both these problems in two ways. Either |
322422de | 239 | transfer and store numbers always in text format, instead of raw |
2c044526 KW |
240 | binary, or else consider using modules like C<Data::Dumper> and |
241 | C<Storable> | |
242 | (included as of Perl 5.8). Keeping all data as text significantly | |
1bcbdd38 | 243 | simplifies matters. |
322422de | 244 | |
433acd8a | 245 | =head2 Files and Filesystems |
e41182b5 GS |
246 | |
247 | Most platforms these days structure files in a hierarchical fashion. | |
b7df3edc | 248 | So, it is reasonably safe to assume that all platforms support the |
6ab3f9cb | 249 | notion of a "path" to uniquely identify a file on the system. How |
b7df3edc | 250 | that path is really written, though, differs considerably. |
e41182b5 | 251 | |
4375e838 | 252 | Although similar, file path specifications differ between Unix, |
b7df3edc GS |
253 | Windows, S<Mac OS>, OS/2, VMS, VOS, S<RISC OS>, and probably others. |
254 | Unix, for example, is one of the few OSes that has the elegant idea | |
255 | of a single root directory. | |
322422de | 256 | |
6ab3f9cb GS |
257 | DOS, OS/2, VMS, VOS, and Windows can work similarly to Unix with C</> |
258 | as path separator, or in their own idiosyncratic ways (such as having | |
259 | several root directories and various "unrooted" device files such NIL: | |
260 | and LPT:). | |
322422de | 261 | |
204ad8d5 | 262 | S<Mac OS> 9 and earlier used C<:> as a path separator instead of C</>. |
322422de | 263 | |
6ab3f9cb GS |
264 | The filesystem may support neither hard links (C<link>) nor |
265 | symbolic links (C<symlink>, C<readlink>, C<lstat>). | |
433acd8a | 266 | |
6ab3f9cb | 267 | The filesystem may support neither access timestamp nor change |
433acd8a JH |
268 | timestamp (meaning that about the only portable timestamp is the |
269 | modification timestamp), or one second granularity of any timestamps | |
270 | (e.g. the FAT filesystem limits the time granularity to two seconds). | |
271 | ||
74555b7a | 272 | The "inode change timestamp" (the C<-C> filetest) may really be the |
e1020413 | 273 | "creation timestamp" (which it is not in Unix). |
95a3fe12 | 274 | |
495c5fdc PG |
275 | VOS perl can emulate Unix filenames with C</> as path separator. The |
276 | native pathname characters greater-than, less-than, number-sign, and | |
277 | percent-sign are always accepted. | |
278 | ||
6ab3f9cb | 279 | S<RISC OS> perl can emulate Unix filenames with C</> as path |
322422de | 280 | separator, or go native and use C<.> for path separator and C<:> to |
6ab3f9cb | 281 | signal filesystems and disk names. |
e41182b5 | 282 | |
e1020413 | 283 | Don't assume Unix filesystem access semantics: that read, write, |
a1667ba3 | 284 | and execute are all the permissions there are, and even if they exist, |
2c044526 | 285 | that their semantics (for example what do C<"r">, C<"w">, and C<"x"> mean on |
e1020413 | 286 | a directory) are the Unix ones. The various Unix/POSIX compatibility |
2c044526 | 287 | layers usually try to make interfaces like C<chmod()> work, but sometimes |
a1667ba3 JH |
288 | there simply is no good mapping. |
289 | ||
35a328a7 CB |
290 | The C<File::Spec> modules provide methods to manipulate path |
291 | specifications and return the results in native format for each | |
292 | platform. This is often unnecessary as Unix-style paths are | |
293 | understood by Perl on every supported platform, but if you need to | |
294 | produce native paths for a native utility that does not understand | |
295 | Unix syntax, or if you are operating on paths or path components | |
296 | in unknown (and thus possibly native) syntax, C<File::Spec> is | |
297 | your friend. Here are two brief examples: | |
e41182b5 | 298 | |
6ab3f9cb GS |
299 | use File::Spec::Functions; |
300 | chdir(updir()); # go up one directory | |
35a328a7 CB |
301 | |
302 | # Concatenate a path from its components | |
303 | my $file = catfile(updir(), 'temp', 'file.txt'); | |
304 | # on Unix: '../temp/file.txt' | |
305 | # on Win32: '..\temp\file.txt' | |
306 | # on VMS: '[-.temp]file.txt' | |
e41182b5 | 307 | |
b7df3edc GS |
308 | In general, production code should not have file paths hardcoded. |
309 | Making them user-supplied or read from a configuration file is | |
310 | better, keeping in mind that file path syntax varies on different | |
311 | machines. | |
e41182b5 GS |
312 | |
313 | This is especially noticeable in scripts like Makefiles and test suites, | |
314 | which often assume C</> as a path separator for subdirectories. | |
315 | ||
2c044526 | 316 | Also of use is C<File::Basename> from the standard distribution, which |
e41182b5 GS |
317 | splits a pathname into pieces (base filename, full path to directory, |
318 | and file suffix). | |
319 | ||
19799a22 | 320 | Even when on a single platform (if you can call Unix a single platform), |
b7df3edc | 321 | remember not to count on the existence or the contents of particular |
3c075c7d | 322 | system-specific files or directories, like F</etc/passwd>, |
b7df3edc GS |
323 | F</etc/sendmail.conf>, F</etc/resolv.conf>, or even F</tmp/>. For |
324 | example, F</etc/passwd> may exist but not contain the encrypted | |
325 | passwords, because the system is using some form of enhanced security. | |
2c044526 | 326 | Or it may not contain all the accounts, because the system is using NIS. |
3c075c7d | 327 | If code does need to rely on such a file, include a description of the |
b7df3edc | 328 | file and its format in the code's documentation, then make it easy for |
3c075c7d CN |
329 | the user to override the default location of the file. |
330 | ||
b7df3edc GS |
331 | Don't assume a text file will end with a newline. They should, |
332 | but people forget. | |
e41182b5 | 333 | |
ec481373 JH |
334 | Do not have two files or directories of the same name with different |
335 | case, like F<test.pl> and F<Test.pl>, as many platforms have | |
336 | case-insensitive (or at least case-forgiving) filenames. Also, try | |
337 | not to have non-word characters (except for C<.>) in the names, and | |
338 | keep them to the 8.3 convention, for maximum portability, onerous a | |
339 | burden though this may appear. | |
dd9f0070 | 340 | |
2c044526 | 341 | Likewise, when using the C<AutoSplit> module, try to keep your functions to |
b7df3edc | 342 | 8.3 naming and case-insensitive conventions; or, at the least, |
dd9f0070 CN |
343 | make it so the resulting files have a unique (case-insensitively) |
344 | first 8 characters. | |
345 | ||
ec481373 JH |
346 | Whitespace in filenames is tolerated on most systems, but not all, |
347 | and even on systems where it might be tolerated, some utilities | |
fe829689 | 348 | might become confused by such whitespace. |
ec481373 | 349 | |
016930a6 JM |
350 | Many systems (DOS, VMS ODS-2) cannot have more than one C<.> in their |
351 | filenames. | |
433acd8a | 352 | |
c47ff5f1 | 353 | Don't assume C<< > >> won't be the first character of a filename. |
fe829689 | 354 | Always use C<< < >> explicitly to open a file for reading, or even |
2c044526 | 355 | better, use the three-arg version of C<open>, unless you want the user to |
fe829689 | 356 | be able to specify a pipe open. |
0a47030a | 357 | |
ceaffd1d | 358 | open my $fh, '<', $existing_file) or die $!; |
0a47030a | 359 | |
6ab3f9cb GS |
360 | If filenames might use strange characters, it is safest to open it |
361 | with C<sysopen> instead of C<open>. C<open> is magic and can | |
c47ff5f1 | 362 | translate characters like C<< > >>, C<< < >>, and C<|>, which may |
b7df3edc | 363 | be the wrong thing to do. (Sometimes, though, it's the right thing.) |
fe829689 PN |
364 | Three-arg open can also help protect against this translation in cases |
365 | where it is undesirable. | |
e41182b5 | 366 | |
ec481373 | 367 | Don't use C<:> as a part of a filename since many systems use that for |
8939ba94 | 368 | their own semantics (Mac OS Classic for separating pathname components, |
ec481373 | 369 | many networking schemes and utilities for separating the nodename and |
08fef530 JH |
370 | the pathname, and so on). For the same reasons, avoid C<@>, C<;> and |
371 | C<|>. | |
ec481373 | 372 | |
e1516da7 JH |
373 | Don't assume that in pathnames you can collapse two leading slashes |
374 | C<//> into one: some networking and clustering filesystems have special | |
2c044526 | 375 | semantics for that. Let the operating system sort it out. |
e1516da7 | 376 | |
ec481373 JH |
377 | The I<portable filename characters> as defined by ANSI C are |
378 | ||
1802421e JK |
379 | 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 |
380 | 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 |
381 | 0 1 2 3 4 5 6 7 8 9 |
382 | . _ - | |
383 | ||
2c044526 | 384 | and the C<"-"> shouldn't be the first character. If you want to be |
e1516da7 JH |
385 | hypercorrect, stay case-insensitive and within the 8.3 naming |
386 | convention (all the files and directories have to be unique within one | |
387 | directory if their names are lowercased and truncated to eight | |
388 | characters before the C<.>, if any, and to three characters after the | |
389 | C<.>, if any). (And do not use C<.>s in directory names.) | |
ec481373 | 390 | |
e41182b5 GS |
391 | =head2 System Interaction |
392 | ||
b7df3edc GS |
393 | Not all platforms provide a command line. These are usually platforms |
394 | that rely primarily on a Graphical User Interface (GUI) for user | |
395 | interaction. A program requiring a command line interface might | |
396 | not work everywhere. This is probably for the user of the program | |
397 | to deal with, so don't stay up late worrying about it. | |
e41182b5 | 398 | |
c87488a3 HS |
399 | Some platforms can't delete or rename files held open by the system, |
400 | this limitation may also apply to changing filesystem metainformation | |
401 | like file permissions or owners. Remember to C<close> files when you | |
402 | are done with them. Don't C<unlink> or C<rename> an open file. Don't | |
403 | C<tie> or C<open> a file already tied or opened; C<untie> or C<close> | |
404 | it first. | |
e41182b5 | 405 | |
0a47030a GS |
406 | Don't open the same file more than once at a time for writing, as some |
407 | operating systems put mandatory locks on such files. | |
408 | ||
73e9292c JH |
409 | Don't assume that write/modify permission on a directory gives the |
410 | right to add or delete files/directories in that directory. That is | |
411 | filesystem specific: in some filesystems you need write/modify | |
412 | permission also (or even just) in the file/directory itself. In some | |
413 | filesystems (AFS, DFS) the permission to add/delete directory entries | |
414 | is a completely separate permission. | |
415 | ||
416 | Don't assume that a single C<unlink> completely gets rid of the file: | |
417 | some filesystems (most notably the ones in VMS) have versioned | |
2c044526 | 418 | filesystems, and C<unlink()> removes only the most recent one (it doesn't |
73e9292c JH |
419 | remove all the versions because by default the native tools on those |
420 | platforms remove just the most recent version, too). The portable | |
421 | idiom to remove all the versions of a file is | |
422 | ||
94bb614c | 423 | 1 while unlink "file"; |
73e9292c JH |
424 | |
425 | This will terminate if the file is undeleteable for some reason | |
426 | (protected, not there, and so on). | |
427 | ||
e41182b5 | 428 | Don't count on a specific environment variable existing in C<%ENV>. |
0a47030a | 429 | Don't count on C<%ENV> entries being case-sensitive, or even |
2c044526 | 430 | case-preserving. Don't try to clear C<%ENV> by saying C<%ENV = ();>, or, |
6be8f7a6 | 431 | if you really have to, make it conditional on C<$^O ne 'VMS'> since in |
4a0d0822 JH |
432 | VMS the C<%ENV> table is much more than a per-process key-value string |
433 | table. | |
e41182b5 | 434 | |
2c044526 | 435 | On VMS, some entries in the C<%ENV> hash are dynamically created when |
c73b03b7 | 436 | their key is used on a read if they did not previously exist. The |
9e861032 | 437 | values for C<$ENV{HOME}>, C<$ENV{TERM}>, C<$ENV{PATH}>, and C<$ENV{USER}>, |
c73b03b7 JM |
438 | are known to be dynamically generated. The specific names that are |
439 | dynamically generated may vary with the version of the C library on VMS, | |
9e861032 | 440 | and more may exist than are documented. |
c73b03b7 | 441 | |
2c044526 KW |
442 | On VMS by default, changes to the %ENV hash persist after perl exits. |
443 | Subsequent invocations of perl in the same process can inadvertently | |
9e861032 | 444 | inherit environment settings that were meant to be temporary. |
c73b03b7 | 445 | |
d1e3b762 | 446 | Don't count on signals or C<%SIG> for anything. |
e41182b5 GS |
447 | |
448 | Don't count on filename globbing. Use C<opendir>, C<readdir>, and | |
449 | C<closedir> instead. | |
450 | ||
b8099c3d | 451 | Don't count on per-program environment variables, or per-program current |
dd9f0070 | 452 | directories. |
b8099c3d | 453 | |
c87488a3 | 454 | Don't count on specific values of C<$!>, neither numeric nor |
2c044526 | 455 | especially the string values. Users may switch their locales causing |
c87488a3 HS |
456 | error messages to be translated into their languages. If you can |
457 | trust a POSIXish environment, you can portably use the symbols defined | |
2c044526 | 458 | by the C<Errno> module, like C<ENOENT>. And don't trust on the values of C<$!> |
c87488a3 | 459 | at all except immediately after a failed system call. |
3c075c7d | 460 | |
a10d74f3 PG |
461 | =head2 Command names versus file pathnames |
462 | ||
463 | Don't assume that the name used to invoke a command or program with | |
464 | C<system> or C<exec> can also be used to test for the existence of the | |
465 | file that holds the executable code for that command or program. | |
68fb0eb7 PG |
466 | First, many systems have "internal" commands that are built-in to the |
467 | shell or OS and while these commands can be invoked, there is no | |
468 | corresponding file. Second, some operating systems (e.g., Cygwin, | |
469 | DJGPP, OS/2, and VOS) have required suffixes for executable files; | |
470 | these suffixes are generally permitted on the command name but are not | |
2c044526 KW |
471 | required. Thus, a command like F<"perl"> might exist in a file named |
472 | F<"perl">, F<"perl.exe">, or F<"perl.pm">, depending on the operating system. | |
473 | The variable C<"_exe"> in the C<Config> module holds the executable suffix, | |
474 | if any. Third, the VMS port carefully sets up C<$^X> and | |
475 | C<$Config{perlpath}> so that no further processing is required. This is | |
68fb0eb7 PG |
476 | just as well, because the matching regular expression used below would |
477 | then have to deal with a possible trailing version number in the VMS | |
478 | file name. | |
a10d74f3 | 479 | |
2c044526 | 480 | To convert C<$^X> to a file pathname, taking account of the requirements |
a10d74f3 | 481 | of the various operating system possibilities, say: |
7ee27b7c | 482 | |
a61fc69c KW |
483 | use Config; |
484 | my $thisperl = $^X; | |
485 | if ($^O ne 'VMS') | |
486 | {$thisperl .= $Config{_exe} unless $thisperl =~ m/$Config{_exe}$/i;} | |
a10d74f3 | 487 | |
2c044526 | 488 | To convert C<$Config{perlpath}> to a file pathname, say: |
7ee27b7c | 489 | |
a61fc69c KW |
490 | use Config; |
491 | my $thisperl = $Config{perlpath}; | |
492 | if ($^O ne 'VMS') | |
493 | {$thisperl .= $Config{_exe} unless $thisperl =~ m/$Config{_exe}$/i;} | |
a10d74f3 | 494 | |
7137b697 JH |
495 | =head2 Networking |
496 | ||
497 | Don't assume that you can reach the public Internet. | |
498 | ||
499 | Don't assume that there is only one way to get through firewalls | |
500 | to the public Internet. | |
501 | ||
932f293e JH |
502 | Don't assume that you can reach outside world through any other port |
503 | than 80, or some web proxy. ftp is blocked by many firewalls. | |
504 | ||
dbc6a9ce JH |
505 | Don't assume that you can send email by connecting to the local SMTP port. |
506 | ||
7137b697 | 507 | Don't assume that you can reach yourself or any node by the name |
dbc6a9ce | 508 | 'localhost'. The same goes for '127.0.0.1'. You will have to try both. |
932f293e | 509 | |
86feb2c5 JH |
510 | Don't assume that the host has only one network card, or that it |
511 | can't bind to many virtual IP addresses. | |
932f293e JH |
512 | |
513 | Don't assume a particular network device name. | |
7137b697 | 514 | |
2c044526 | 515 | Don't assume a particular set of C<ioctl()>s will work. |
7137b697 JH |
516 | |
517 | Don't assume that you can ping hosts and get replies. | |
518 | ||
dbc6a9ce JH |
519 | Don't assume that any particular port (service) will respond. |
520 | ||
2c044526 | 521 | Don't assume that C<Sys::Hostname> (or any other API or command) returns |
ac036724 | 522 | either a fully qualified hostname or a non-qualified hostname: it all |
523 | depends on how the system had been configured. Also remember that for | |
524 | things such as DHCP and NAT, the hostname you get back might not be | |
525 | very useful. | |
dbc6a9ce | 526 | |
a746ef5c | 527 | All the above I<don't>s may look daunting, and they are, but the key |
932f293e JH |
528 | is to degrade gracefully if one cannot reach the particular network |
529 | service one wants. Croaking or hanging do not look very professional. | |
530 | ||
e41182b5 GS |
531 | =head2 Interprocess Communication (IPC) |
532 | ||
b7df3edc GS |
533 | In general, don't directly access the system in code meant to be |
534 | portable. That means, no C<system>, C<exec>, C<fork>, C<pipe>, | |
535 | C<``>, C<qx//>, C<open> with a C<|>, nor any of the other things | |
2c044526 | 536 | that makes being a Perl hacker worth being. |
e41182b5 GS |
537 | |
538 | Commands that launch external processes are generally supported on | |
b7df3edc GS |
539 | most platforms (though many of them do not support any type of |
540 | forking). The problem with using them arises from what you invoke | |
541 | them on. External tools are often named differently on different | |
4375e838 | 542 | platforms, may not be available in the same location, might accept |
b7df3edc GS |
543 | different arguments, can behave differently, and often present their |
544 | results in a platform-dependent way. Thus, you should seldom depend | |
2890cc8c | 545 | on them to produce consistent results. (Then again, if you're calling |
b7df3edc | 546 | I<netstat -a>, you probably don't expect it to run on both Unix and CP/M.) |
e41182b5 | 547 | |
b7df3edc | 548 | One especially common bit of Perl code is opening a pipe to B<sendmail>: |
e41182b5 | 549 | |
2890cc8c | 550 | open(MAIL, '|/usr/lib/sendmail -t') |
b7df3edc | 551 | or die "cannot fork sendmail: $!"; |
e41182b5 GS |
552 | |
553 | This is fine for systems programming when sendmail is known to be | |
554 | available. But it is not fine for many non-Unix systems, and even | |
555 | some Unix systems that may not have sendmail installed. If a portable | |
b7df3edc | 556 | solution is needed, see the various distributions on CPAN that deal |
2c044526 KW |
557 | with it. C<Mail::Mailer> and C<Mail::Send> in the C<MailTools> distribution are |
558 | commonly used, and provide several mailing methods, including C<mail>, | |
559 | C<sendmail>, and direct SMTP (via C<Net::SMTP>) if a mail transfer agent is | |
560 | not available. C<Mail::Sendmail> is a standalone module that provides | |
b7df3edc GS |
561 | simple, platform-independent mailing. |
562 | ||
563 | The Unix System V IPC (C<msg*(), sem*(), shm*()>) is not available | |
564 | even on all Unix platforms. | |
e41182b5 | 565 | |
a81e5e2e A |
566 | Do not use either the bare result of C<pack("N", 10, 20, 30, 40)> or |
567 | bare v-strings (such as C<v10.20.30.40>) to represent IPv4 addresses: | |
568 | both forms just pack the four bytes into network order. That this | |
569 | would be equal to the C language C<in_addr> struct (which is what the | |
570 | socket code internally uses) is not guaranteed. To be portable use | |
2c044526 | 571 | the routines of the C<Socket> extension, such as C<inet_aton()>, |
a81e5e2e | 572 | C<inet_ntoa()>, and C<sockaddr_in()>. |
6b2463a0 | 573 | |
e41182b5 | 574 | The rule of thumb for portable code is: Do it all in portable Perl, or |
0a47030a | 575 | use a module (that may internally implement it with platform-specific |
2c044526 | 576 | code, but exposes a common interface). |
e41182b5 | 577 | |
e41182b5 GS |
578 | =head2 External Subroutines (XS) |
579 | ||
b7df3edc | 580 | XS code can usually be made to work with any platform, but dependent |
e41182b5 GS |
581 | libraries, header files, etc., might not be readily available or |
582 | portable, or the XS code itself might be platform-specific, just as Perl | |
583 | code might be. If the libraries and headers are portable, then it is | |
584 | normally reasonable to make sure the XS code is portable, too. | |
585 | ||
b7df3edc GS |
586 | A different type of portability issue arises when writing XS code: |
587 | availability of a C compiler on the end-user's system. C brings | |
588 | with it its own portability issues, and writing XS code will expose | |
589 | you to some of those. Writing purely in Perl is an easier way to | |
e41182b5 GS |
590 | achieve portability. |
591 | ||
e41182b5 GS |
592 | =head2 Standard Modules |
593 | ||
594 | In general, the standard modules work across platforms. Notable | |
2c044526 | 595 | exceptions are the C<CPAN> module (which currently makes connections to external |
e41182b5 | 596 | programs that may not be available), platform-specific modules (like |
2c044526 | 597 | C<ExtUtils::MM_VMS>), and DBM modules. |
e41182b5 | 598 | |
b7df3edc | 599 | There is no one DBM module available on all platforms. |
2c044526 KW |
600 | C<SDBM_File> and the others are generally available on all Unix and DOSish |
601 | ports, but not in MacPerl, where only C<NDBM_File> and C<DB_File> are | |
0a47030a | 602 | available. |
e41182b5 GS |
603 | |
604 | The good news is that at least some DBM module should be available, and | |
2c044526 | 605 | C<AnyDBM_File> will use whichever module it can find. Of course, then |
b7df3edc GS |
606 | the code needs to be fairly strict, dropping to the greatest common |
607 | factor (e.g., not exceeding 1K for each record), so that it will | |
6ab3f9cb | 608 | work with any DBM module. See L<AnyDBM_File> for more details. |
e41182b5 | 609 | |
e41182b5 GS |
610 | =head2 Time and Date |
611 | ||
0a47030a | 612 | The system's notion of time of day and calendar date is controlled in |
b7df3edc | 613 | widely different ways. Don't assume the timezone is stored in C<$ENV{TZ}>, |
0a47030a | 614 | and even if it is, don't assume that you can control the timezone through |
c87488a3 HS |
615 | that variable. Don't assume anything about the three-letter timezone |
616 | abbreviations (for example that MST would be the Mountain Standard Time, | |
617 | it's been known to stand for Moscow Standard Time). If you need to | |
618 | use timezones, express them in some unambiguous format like the | |
619 | exact number of minutes offset from UTC, or the POSIX timezone | |
620 | format. | |
e41182b5 | 621 | |
322422de | 622 | Don't assume that the epoch starts at 00:00:00, January 1, 1970, |
c87488a3 HS |
623 | because that is OS- and implementation-specific. It is better to |
624 | store a date in an unambiguous representation. The ISO 8601 standard | |
766af94f | 625 | defines YYYY-MM-DD as the date format, or YYYY-MM-DDTHH:MM:SS |
c87488a3 | 626 | (that's a literal "T" separating the date from the time). |
91d20606 | 627 | Please do use the ISO 8601 instead of making us guess what |
c87488a3 HS |
628 | date 02/03/04 might be. ISO 8601 even sorts nicely as-is. |
629 | A text representation (like "1987-12-18") can be easily converted | |
2c044526 | 630 | into an OS-specific value using a module like C<Date::Parse>. |
c87488a3 | 631 | An array of values, such as those returned by C<localtime>, can be |
2c044526 | 632 | converted to an OS-specific representation using C<Time::Local>. |
322422de | 633 | |
19799a22 GS |
634 | When calculating specific times, such as for tests in time or date modules, |
635 | it may be appropriate to calculate an offset for the epoch. | |
b7df3edc | 636 | |
19799a22 | 637 | require Time::Local; |
ceaffd1d | 638 | my $offset = Time::Local::timegm(0, 0, 0, 1, 0, 70); |
b7df3edc | 639 | |
204ad8d5 JV |
640 | The value for C<$offset> in Unix will be C<0>, but in Mac OS Classic |
641 | will be some large number. C<$offset> can then be added to a Unix time | |
642 | value to get what should be the proper value on any system. | |
322422de GS |
643 | |
644 | =head2 Character sets and character encoding | |
645 | ||
ec481373 JH |
646 | Assume very little about character sets. |
647 | ||
648 | Assume nothing about numerical values (C<ord>, C<chr>) of characters. | |
eb9df707 KW |
649 | Do not use explicit code point ranges (like C<\xHH-\xHH)>. However, |
650 | starting in Perl v5.22, regular expression pattern bracketed character | |
f4240379 KW |
651 | class ranges specified like C<qr/[\N{U+HH}-\N{U+HH}]/> are portable, |
652 | and starting in Perl v5.24, the same ranges are portable in C<tr///>. | |
2c044526 | 653 | You can portably use symbolic character classes like C<[:print:]>. |
ec481373 JH |
654 | |
655 | Do not assume that the alphabetic characters are encoded contiguously | |
eb9df707 KW |
656 | (in the numeric sense). There may be gaps. Special coding in Perl, |
657 | however, guarantees that all subsets of C<qr/[A-Z]/>, C<qr/[a-z]/>, and | |
2c044526 KW |
658 | C<qr/[0-9]/> behave as expected. C<tr///> behaves the same for these |
659 | ranges. In patterns, any ranges specified with end points using the | |
660 | C<\N{...}> notations ensures character set portability, but it is a bug | |
f4240379 | 661 | in Perl v5.22, that this isn't true of C<tr///>, fixed in v5.24. |
ec481373 JH |
662 | |
663 | Do not assume anything about the ordering of the characters. | |
664 | The lowercase letters may come before or after the uppercase letters; | |
b432a672 AL |
665 | the lowercase and uppercase may be interlaced so that both "a" and "A" |
666 | come before "b"; the accented and other international characters may | |
667 | be interlaced so that E<auml> comes before "b". | |
2c044526 | 668 | L<Unicode::Collate> can be used to sort this all out. |
322422de GS |
669 | |
670 | =head2 Internationalisation | |
671 | ||
b7df3edc GS |
672 | If you may assume POSIX (a rather large assumption), you may read |
673 | more about the POSIX locale system from L<perllocale>. The locale | |
674 | system at least attempts to make things a little bit more portable, | |
675 | or at least more convenient and native-friendly for non-English | |
676 | users. The system affects character sets and encoding, and date | |
677 | and time formatting--amongst other things. | |
e41182b5 | 678 | |
c87488a3 HS |
679 | If you really want to be international, you should consider Unicode. |
680 | See L<perluniintro> and L<perlunicode> for more information. | |
681 | ||
11264fdb JH |
682 | If you want to use non-ASCII bytes (outside the bytes 0x00..0x7f) in |
683 | the "source code" of your code, to be portable you have to be explicit | |
684 | about what bytes they are. Someone might for example be using your | |
685 | code under a UTF-8 locale, in which case random native bytes might be | |
686 | illegal ("Malformed UTF-8 ...") This means that for example embedding | |
687 | ISO 8859-1 bytes beyond 0x7f into your strings might cause trouble | |
688 | later. If the bytes are native 8-bit bytes, you can use the C<bytes> | |
eb9df707 KW |
689 | pragma. If the bytes are in a string (regular expressions being |
690 | curious strings), you can often also use the C<\xHH> or more portably, | |
691 | the C<\N{U+HH}> notations instead | |
1bcbdd38 | 692 | of embedding the bytes as-is. If you want to write your code in UTF-8, |
eb9df707 | 693 | you can use L<utf8>. |
11264fdb | 694 | |
e41182b5 GS |
695 | =head2 System Resources |
696 | ||
0a47030a GS |
697 | If your code is destined for systems with severely constrained (or |
698 | missing!) virtual memory systems then you want to be I<especially> mindful | |
699 | of avoiding wasteful constructs such as: | |
e41182b5 | 700 | |
ceaffd1d | 701 | my @lines = <$very_large_file>; # bad |
e41182b5 | 702 | |
ceaffd1d | 703 | while (<$fh>) {$file .= $_} # sometimes bad |
704 | my $file = join('', <$fh>); # better | |
e41182b5 | 705 | |
b7df3edc GS |
706 | The last two constructs may appear unintuitive to most people. The |
707 | first repeatedly grows a string, whereas the second allocates a | |
708 | large chunk of memory in one go. On some systems, the second is | |
2c044526 | 709 | more efficient than the first. |
0a47030a | 710 | |
e41182b5 GS |
711 | =head2 Security |
712 | ||
b7df3edc | 713 | Most multi-user platforms provide basic levels of security, usually |
ac036724 | 714 | implemented at the filesystem level. Some, however, unfortunately do |
715 | not. Thus the notion of user id, or "home" directory, | |
b7df3edc GS |
716 | or even the state of being logged-in, may be unrecognizable on many |
717 | platforms. If you write programs that are security-conscious, it | |
718 | is usually best to know what type of system you will be running | |
719 | under so that you can write code explicitly for that platform (or | |
720 | class of platforms). | |
0a47030a | 721 | |
e1020413 | 722 | Don't assume the Unix filesystem access semantics: the operating |
a1667ba3 | 723 | system or the filesystem may be using some ACL systems, which are |
2c044526 | 724 | richer languages than the usual C<rwx>. Even if the C<rwx> exist, |
a1667ba3 JH |
725 | their semantics might be different. |
726 | ||
2c044526 | 727 | (From the security viewpoint, testing for permissions before attempting to |
a1667ba3 | 728 | do something is silly anyway: if one tries this, there is potential |
ac036724 | 729 | for race conditions. Someone or something might change the |
a1667ba3 JH |
730 | permissions between the permissions check and the actual operation. |
731 | Just try the operation.) | |
732 | ||
e1020413 | 733 | Don't assume the Unix user and group semantics: especially, don't |
2c044526 | 734 | expect C<< $< >> and C<< $> >> (or C<$(> and C<$)>) to work |
a1667ba3 JH |
735 | for switching identities (or memberships). |
736 | ||
737 | Don't assume set-uid and set-gid semantics. (And even if you do, | |
738 | think twice: set-uid and set-gid are a known can of security worms.) | |
739 | ||
e41182b5 GS |
740 | =head2 Style |
741 | ||
742 | For those times when it is necessary to have platform-specific code, | |
743 | consider keeping the platform-specific code in one place, making porting | |
2c044526 | 744 | to other platforms easier. Use the C<Config> module and the special |
0a47030a GS |
745 | variable C<$^O> to differentiate platforms, as described in |
746 | L<"PLATFORMS">. | |
e41182b5 | 747 | |
d4c800c7 JH |
748 | Beware of the "else syndrome": |
749 | ||
750 | if ($^O eq 'MSWin32') { | |
751 | # code that assumes Windows | |
752 | } else { | |
753 | # code that assumes Linux | |
754 | } | |
755 | ||
756 | The C<else> branch should be used for the really ultimate fallback, | |
757 | not for code specific to some platform. | |
758 | ||
b7df3edc GS |
759 | Be careful in the tests you supply with your module or programs. |
760 | Module code may be fully portable, but its tests might not be. This | |
761 | often happens when tests spawn off other processes or call external | |
762 | programs to aid in the testing, or when (as noted above) the tests | |
c87488a3 HS |
763 | assume certain things about the filesystem and paths. Be careful not |
764 | to depend on a specific output style for errors, such as when checking | |
765 | C<$!> after a failed system call. Using C<$!> for anything else than | |
2c044526 | 766 | displaying it as output is doubtful (though see the C<Errno> module for |
c87488a3 HS |
767 | testing reasonably portably for error value). Some platforms expect |
768 | a certain output format, and Perl on those platforms may have been | |
769 | adjusted accordingly. Most specifically, don't anchor a regex when | |
770 | testing an error value. | |
e41182b5 | 771 | |
0a47030a | 772 | =head1 CPAN Testers |
e41182b5 | 773 | |
0a47030a GS |
774 | Modules uploaded to CPAN are tested by a variety of volunteers on |
775 | different platforms. These CPAN testers are notified by mail of each | |
e41182b5 | 776 | new upload, and reply to the list with PASS, FAIL, NA (not applicable to |
0a47030a | 777 | this platform), or UNKNOWN (unknown), along with any relevant notations. |
e41182b5 GS |
778 | |
779 | The purpose of the testing is twofold: one, to help developers fix any | |
0a47030a | 780 | problems in their code that crop up because of lack of testing on other |
b7df3edc | 781 | platforms; two, to provide users with information about whether |
0a47030a | 782 | a given module works on a given platform. |
e41182b5 | 783 | |
2890cc8c | 784 | Also see: |
7ee27b7c | 785 | |
e41182b5 GS |
786 | =over 4 |
787 | ||
7ee27b7c AT |
788 | =item * |
789 | ||
636280bd | 790 | Mailing list: cpan-testers-discuss@perl.org |
7ee27b7c AT |
791 | |
792 | =item * | |
e41182b5 | 793 | |
500f1b69 | 794 | Testing results: L<http://www.cpantesters.org/> |
e41182b5 GS |
795 | |
796 | =back | |
797 | ||
e41182b5 GS |
798 | =head1 PLATFORMS |
799 | ||
1bcbdd38 BF |
800 | Perl is built with a C<$^O> variable that indicates the operating |
801 | system it was built on. This was implemented | |
b7df3edc GS |
802 | to help speed up code that would otherwise have to C<use Config> |
803 | and use the value of C<$Config{osname}>. Of course, to get more | |
e41182b5 GS |
804 | detailed information about the system, looking into C<%Config> is |
805 | certainly recommended. | |
806 | ||
b7df3edc GS |
807 | C<%Config> cannot always be trusted, however, because it was built |
808 | at compile time. If perl was built in one place, then transferred | |
809 | elsewhere, some values may be wrong. The values may even have been | |
810 | edited after the fact. | |
6ab3f9cb | 811 | |
e41182b5 GS |
812 | =head2 Unix |
813 | ||
814 | Perl works on a bewildering variety of Unix and Unix-like platforms (see | |
815 | e.g. most of the files in the F<hints/> directory in the source code kit). | |
816 | On most of these systems, the value of C<$^O> (hence C<$Config{'osname'}>, | |
d1e3b762 GS |
817 | too) is determined either by lowercasing and stripping punctuation from the |
818 | first field of the string returned by typing C<uname -a> (or a similar command) | |
819 | at the shell prompt or by testing the file system for the presence of | |
820 | uniquely named files such as a kernel or header file. Here, for example, | |
821 | are a few of the more popular Unix flavors: | |
e41182b5 | 822 | |
b7df3edc | 823 | uname $^O $Config{'archname'} |
6ab3f9cb | 824 | -------------------------------------------- |
b7df3edc | 825 | AIX aix aix |
6ab3f9cb | 826 | BSD/OS bsdos i386-bsdos |
e1516da7 | 827 | Darwin darwin darwin |
6ab3f9cb | 828 | DYNIX/ptx dynixptx i386-dynixptx |
2890cc8c | 829 | FreeBSD freebsd freebsd-i386 |
df00ff3b | 830 | Haiku haiku BePC-haiku |
d1e3b762 | 831 | Linux linux arm-linux |
56b575b9 | 832 | Linux linux armv5tel-linux |
b7df3edc | 833 | Linux linux i386-linux |
6ab3f9cb GS |
834 | Linux linux i586-linux |
835 | Linux linux ppc-linux | |
b7df3edc GS |
836 | HP-UX hpux PA-RISC1.1 |
837 | IRIX irix irix | |
b787fad4 | 838 | Mac OS X darwin darwin |
d1e3b762 GS |
839 | NeXT 3 next next-fat |
840 | NeXT 4 next OPENSTEP-Mach | |
6ab3f9cb | 841 | openbsd openbsd i386-openbsd |
b7df3edc | 842 | OSF1 dec_osf alpha-dec_osf |
6ab3f9cb GS |
843 | reliantunix-n svr4 RM400-svr4 |
844 | SCO_SV sco_sv i386-sco_sv | |
845 | SINIX-N svr4 RM400-svr4 | |
846 | sn4609 unicos CRAY_C90-unicos | |
847 | sn6521 unicosmk t3e-unicosmk | |
848 | sn9617 unicos CRAY_J90-unicos | |
b7df3edc GS |
849 | SunOS solaris sun4-solaris |
850 | SunOS solaris i86pc-solaris | |
851 | SunOS4 sunos sun4-sunos | |
e41182b5 | 852 | |
b7df3edc GS |
853 | Because the value of C<$Config{archname}> may depend on the |
854 | hardware architecture, it can vary more than the value of C<$^O>. | |
6ab3f9cb | 855 | |
e41182b5 GS |
856 | =head2 DOS and Derivatives |
857 | ||
b7df3edc | 858 | Perl has long been ported to Intel-style microcomputers running under |
e41182b5 GS |
859 | systems like PC-DOS, MS-DOS, OS/2, and most Windows platforms you can |
860 | bring yourself to mention (except for Windows CE, if you count that). | |
b7df3edc | 861 | Users familiar with I<COMMAND.COM> or I<CMD.EXE> style shells should |
e41182b5 GS |
862 | be aware that each of these file specifications may have subtle |
863 | differences: | |
864 | ||
ceaffd1d | 865 | my $filespec0 = "c:/foo/bar/file.txt"; |
866 | my $filespec1 = "c:\\foo\\bar\\file.txt"; | |
867 | my $filespec2 = 'c:\foo\bar\file.txt'; | |
868 | my $filespec3 = 'c:\\foo\\bar\\file.txt'; | |
e41182b5 | 869 | |
b7df3edc GS |
870 | System calls accept either C</> or C<\> as the path separator. |
871 | However, many command-line utilities of DOS vintage treat C</> as | |
872 | the option prefix, so may get confused by filenames containing C</>. | |
873 | Aside from calling any external programs, C</> will work just fine, | |
874 | and probably better, as it is more consistent with popular usage, | |
875 | and avoids the problem of remembering what to backwhack and what | |
876 | not to. | |
e41182b5 | 877 | |
b7df3edc GS |
878 | The DOS FAT filesystem can accommodate only "8.3" style filenames. Under |
879 | the "case-insensitive, but case-preserving" HPFS (OS/2) and NTFS (NT) | |
0a47030a | 880 | filesystems you may have to be careful about case returned with functions |
e41182b5 GS |
881 | like C<readdir> or used with functions like C<open> or C<opendir>. |
882 | ||
b7df3edc GS |
883 | DOS also treats several filenames as special, such as AUX, PRN, |
884 | NUL, CON, COM1, LPT1, LPT2, etc. Unfortunately, sometimes these | |
885 | filenames won't even work if you include an explicit directory | |
886 | prefix. It is best to avoid such filenames, if you want your code | |
887 | to be portable to DOS and its derivatives. It's hard to know what | |
888 | these all are, unfortunately. | |
e41182b5 GS |
889 | |
890 | Users of these operating systems may also wish to make use of | |
b7df3edc | 891 | scripts such as I<pl2bat.bat> or I<pl2cmd> to |
e41182b5 GS |
892 | put wrappers around your scripts. |
893 | ||
894 | Newline (C<\n>) is translated as C<\015\012> by STDIO when reading from | |
6ab3f9cb GS |
895 | and writing to files (see L<"Newlines">). C<binmode(FILEHANDLE)> |
896 | will keep C<\n> translated as C<\012> for that filehandle. Since it is a | |
897 | no-op on other systems, C<binmode> should be used for cross-platform code | |
b7df3edc GS |
898 | that deals with binary data. That's assuming you realize in advance |
899 | that your data is in binary. General-purpose programs should | |
900 | often assume nothing about their data. | |
e41182b5 | 901 | |
b7df3edc | 902 | The C<$^O> variable and the C<$Config{archname}> values for various |
e41182b5 GS |
903 | DOSish perls are as follows: |
904 | ||
67ac489e MS |
905 | OS $^O $Config{archname} ID Version |
906 | -------------------------------------------------------- | |
2890cc8c CBW |
907 | MS-DOS dos ? |
908 | PC-DOS dos ? | |
67ac489e MS |
909 | OS/2 os2 ? |
910 | Windows 3.1 ? ? 0 3 01 | |
911 | Windows 95 MSWin32 MSWin32-x86 1 4 00 | |
912 | Windows 98 MSWin32 MSWin32-x86 1 4 10 | |
913 | Windows ME MSWin32 MSWin32-x86 1 ? | |
914 | Windows NT MSWin32 MSWin32-x86 2 4 xx | |
915 | Windows NT MSWin32 MSWin32-ALPHA 2 4 xx | |
916 | Windows NT MSWin32 MSWin32-ppc 2 4 xx | |
7ee27b7c AT |
917 | Windows 2000 MSWin32 MSWin32-x86 2 5 00 |
918 | Windows XP MSWin32 MSWin32-x86 2 5 01 | |
919 | Windows 2003 MSWin32 MSWin32-x86 2 5 02 | |
bc643a33 CJ |
920 | Windows Vista MSWin32 MSWin32-x86 2 6 00 |
921 | Windows 7 MSWin32 MSWin32-x86 2 6 01 | |
922 | Windows 7 MSWin32 MSWin32-x64 2 6 01 | |
ae525322 CBW |
923 | Windows 2008 MSWin32 MSWin32-x86 2 6 01 |
924 | Windows 2008 MSWin32 MSWin32-x64 2 6 01 | |
2890cc8c | 925 | Windows CE MSWin32 ? 3 |
7ee27b7c | 926 | Cygwin cygwin cygwin |
e41182b5 | 927 | |
34aaaa84 | 928 | The various MSWin32 Perl's can distinguish the OS they are running on |
2890cc8c | 929 | via the value of the fifth element of the list returned from |
2c044526 | 930 | C<Win32::GetOSVersion()>. For example: |
34aaaa84 PP |
931 | |
932 | if ($^O eq 'MSWin32') { | |
933 | my @os_version_info = Win32::GetOSVersion(); | |
934 | print +('3.1','95','NT')[$os_version_info[4]],"\n"; | |
935 | } | |
936 | ||
2c044526 | 937 | There are also C<Win32::IsWinNT()> and C<Win32::IsWin95()>; try C<perldoc Win32>, |
7939d86b | 938 | and as of libwin32 0.19 (not part of the core Perl distribution) |
2c044526 | 939 | C<Win32::GetOSName()>. The very portable C<POSIX::uname()> will work too: |
1d65be3a JH |
940 | |
941 | c:\> perl -MPOSIX -we "print join '|', uname" | |
942 | Windows NT|moonru|5.0|Build 2195 (Service Pack 2)|x86 | |
d99f392e | 943 | |
e41182b5 GS |
944 | Also see: |
945 | ||
946 | =over 4 | |
947 | ||
c997b287 | 948 | =item * |
e41182b5 | 949 | |
500f1b69 | 950 | The djgpp environment for DOS, L<http://www.delorie.com/djgpp/> |
c997b287 | 951 | and L<perldos>. |
e41182b5 | 952 | |
c997b287 | 953 | =item * |
e41182b5 | 954 | |
c997b287 | 955 | The EMX environment for DOS, OS/2, etc. emx@iaehv.nl, |
500f1b69 | 956 | L<ftp://hobbes.nmsu.edu/pub/os2/dev/emx/> Also L<perlos2>. |
e41182b5 | 957 | |
c997b287 | 958 | =item * |
d1e3b762 | 959 | |
c997b287 | 960 | Build instructions for Win32 in L<perlwin32>, or under the Cygnus environment |
2890cc8c | 961 | in L<perlcygwin>. |
c997b287 GS |
962 | |
963 | =item * | |
964 | ||
965 | The C<Win32::*> modules in L<Win32>. | |
966 | ||
967 | =item * | |
968 | ||
500f1b69 | 969 | The ActiveState Pages, L<http://www.activestate.com/> |
c997b287 GS |
970 | |
971 | =item * | |
972 | ||
2890cc8c | 973 | The Cygwin environment for Win32; F<README.cygwin> (installed |
500f1b69 | 974 | as L<perlcygwin>), L<http://www.cygwin.com/> |
c997b287 GS |
975 | |
976 | =item * | |
977 | ||
978 | The U/WIN environment for Win32, | |
500f1b69 | 979 | L<http://www.research.att.com/sw/tools/uwin/> |
c997b287 | 980 | |
cea6626f | 981 | =item * |
d1e3b762 | 982 | |
cea6626f | 983 | Build instructions for OS/2, L<perlos2> |
d1e3b762 | 984 | |
e41182b5 GS |
985 | =back |
986 | ||
e41182b5 GS |
987 | =head2 VMS |
988 | ||
2c044526 | 989 | Perl on VMS is discussed in L<perlvms> in the Perl distribution. |
016930a6 JM |
990 | |
991 | The official name of VMS as of this writing is OpenVMS. | |
992 | ||
e41182b5 GS |
993 | Interacting with Perl from the Digital Command Language (DCL) shell |
994 | often requires a different set of quotation marks than Unix shells do. | |
995 | For example: | |
996 | ||
997 | $ perl -e "print ""Hello, world.\n""" | |
998 | Hello, world. | |
999 | ||
2c044526 | 1000 | There are several ways to wrap your Perl scripts in DCL F<.COM> files, if |
e41182b5 GS |
1001 | you are so inclined. For example: |
1002 | ||
1003 | $ write sys$output "Hello from DCL!" | |
1004 | $ if p1 .eqs. "" | |
1005 | $ then perl -x 'f$environment("PROCEDURE") | |
1006 | $ else perl -x - 'p1 'p2 'p3 'p4 'p5 'p6 'p7 'p8 | |
1007 | $ deck/dollars="__END__" | |
1008 | #!/usr/bin/perl | |
1009 | ||
1010 | print "Hello from Perl!\n"; | |
1011 | ||
1012 | __END__ | |
1013 | $ endif | |
1014 | ||
1015 | Do take care with C<$ ASSIGN/nolog/user SYS$COMMAND: SYS$INPUT> if your | |
2c044526 | 1016 | Perl-in-DCL script expects to do things like C<< $read = <STDIN>; >>. |
e41182b5 | 1017 | |
9e861032 CB |
1018 | The VMS operating system has two filesystems, designated by their |
1019 | on-disk structure (ODS) level: ODS-2 and its successor ODS-5. The | |
1020 | initial port of Perl to VMS pre-dates ODS-5, but all current testing and | |
1021 | development assumes ODS-5 and its capabilities, including case | |
1022 | preservation, extended characters in filespecs, and names up to 8192 | |
1023 | bytes long. | |
016930a6 | 1024 | |
9e861032 CB |
1025 | Perl on VMS can accept either VMS- or Unix-style file |
1026 | specifications as in either of the following: | |
1089a9e3 | 1027 | |
9e861032 CB |
1028 | $ perl -ne "print if /perl_setup/i" SYS$LOGIN:LOGIN.COM |
1029 | $ perl -ne "print if /perl_setup/i" /sys$login/login.com | |
1089a9e3 | 1030 | |
9e861032 | 1031 | but not a mixture of both as in: |
1089a9e3 | 1032 | |
9e861032 CB |
1033 | $ perl -ne "print if /perl_setup/i" sys$login:/login.com |
1034 | Can't open sys$login:/login.com: file specification syntax error | |
e41182b5 | 1035 | |
9e861032 CB |
1036 | In general, the easiest path to portability is always to specify |
1037 | filenames in Unix format unless they will need to be processed by native | |
1038 | commands or utilities. Because of this latter consideration, the | |
1039 | File::Spec module by default returns native format specifications | |
1040 | regardless of input format. This default may be reversed so that | |
1041 | filenames are always reported in Unix format by specifying the | |
1042 | C<DECC$FILENAME_UNIX_REPORT> feature logical in the environment. | |
1043 | ||
1044 | The file type, or extension, is always present in a VMS-format file | |
1045 | specification even if it's zero-length. This means that, by default, | |
1046 | C<readdir> will return a trailing dot on a file with no extension, so | |
2c044526 KW |
1047 | where you would see C<"a"> on Unix you'll see C<"a."> on VMS. However, |
1048 | the trailing dot may be suppressed by enabling the | |
9e861032 CB |
1049 | C<DECC$READDIR_DROPDOTNOTYPE> feature in the environment (see the CRTL |
1050 | documentation on feature logical names). | |
e41182b5 | 1051 | |
5e12dbfa | 1052 | What C<\n> represents depends on the type of file opened. It usually |
2890cc8c CBW |
1053 | represents C<\012> but it could also be C<\015>, C<\012>, C<\015\012>, |
1054 | C<\000>, C<\040>, or nothing depending on the file organization and | |
2c044526 KW |
1055 | record format. The C<VMS::Stdio> module provides access to the |
1056 | special C<fopen()> requirements of files with unusual attributes on VMS. | |
e41182b5 | 1057 | |
e41182b5 | 1058 | The value of C<$^O> on OpenVMS is "VMS". To determine the architecture |
9e861032 | 1059 | that you are running on refer to C<$Config{'archname'}>. |
016930a6 | 1060 | |
b7df3edc GS |
1061 | On VMS, perl determines the UTC offset from the C<SYS$TIMEZONE_DIFFERENTIAL> |
1062 | logical name. Although the VMS epoch began at 17-NOV-1858 00:00:00.00, | |
6ab3f9cb | 1063 | calls to C<localtime> are adjusted to count offsets from |
b7df3edc | 1064 | 01-JAN-1970 00:00:00.00, just like Unix. |
6ab3f9cb | 1065 | |
e41182b5 GS |
1066 | Also see: |
1067 | ||
1068 | =over 4 | |
1069 | ||
c997b287 GS |
1070 | =item * |
1071 | ||
96090e4f | 1072 | F<README.vms> (installed as F<README_vms>), L<perlvms> |
c997b287 GS |
1073 | |
1074 | =item * | |
1075 | ||
1089a9e3 | 1076 | vmsperl list, vmsperl-subscribe@perl.org |
e41182b5 | 1077 | |
c997b287 | 1078 | =item * |
e41182b5 | 1079 | |
500f1b69 | 1080 | vmsperl on the web, L<http://www.sidhe.org/vmsperl/index.html> |
e41182b5 | 1081 | |
9e861032 CB |
1082 | =item * |
1083 | ||
1084 | VMS Software Inc. web site, L<http://www.vmssoftware.com> | |
1085 | ||
e41182b5 GS |
1086 | =back |
1087 | ||
495c5fdc PG |
1088 | =head2 VOS |
1089 | ||
10fb90aa | 1090 | Perl on VOS (also known as OpenVOS) is discussed in F<README.vos> |
2c044526 | 1091 | in the Perl distribution (installed as L<perlvos>). Perl on VOS |
10fb90aa PG |
1092 | can accept either VOS- or Unix-style file specifications as in |
1093 | either of the following: | |
495c5fdc | 1094 | |
ea8b8ad2 VP |
1095 | $ perl -ne "print if /perl_setup/i" >system>notices |
1096 | $ perl -ne "print if /perl_setup/i" /system/notices | |
495c5fdc PG |
1097 | |
1098 | or even a mixture of both as in: | |
1099 | ||
ea8b8ad2 | 1100 | $ perl -ne "print if /perl_setup/i" >system/notices |
495c5fdc | 1101 | |
b7df3edc | 1102 | Even though VOS allows the slash character to appear in object |
495c5fdc | 1103 | names, because the VOS port of Perl interprets it as a pathname |
10fb90aa PG |
1104 | delimiting character, VOS files, directories, or links whose |
1105 | names contain a slash character cannot be processed. Such files | |
1106 | must be renamed before they can be processed by Perl. | |
1107 | ||
1108 | Older releases of VOS (prior to OpenVOS Release 17.0) limit file | |
1109 | names to 32 or fewer characters, prohibit file names from | |
1110 | starting with a C<-> character, and prohibit file names from | |
1111 | containing any character matching C<< tr/ !#%&'()*;<=>?// >>. | |
1112 | ||
1113 | Newer releases of VOS (OpenVOS Release 17.0 or later) support a | |
1114 | feature known as extended names. On these releases, file names | |
1115 | can contain up to 255 characters, are prohibited from starting | |
1116 | with a C<-> character, and the set of prohibited characters is | |
1117 | reduced to any character matching C<< tr/#%*<>?// >>. There are | |
c69ca1d4 | 1118 | restrictions involving spaces and apostrophes: these characters |
10fb90aa PG |
1119 | must not begin or end a name, nor can they immediately precede or |
1120 | follow a period. Additionally, a space must not immediately | |
1121 | precede another space or hyphen. Specifically, the following | |
1122 | character combinations are prohibited: space-space, | |
1123 | space-hyphen, period-space, space-period, period-apostrophe, | |
1124 | apostrophe-period, leading or trailing space, and leading or | |
1125 | trailing apostrophe. Although an extended file name is limited | |
1126 | to 255 characters, a path name is still limited to 256 | |
1127 | characters. | |
1128 | ||
7d4dfb6d | 1129 | The value of C<$^O> on VOS is "vos". To determine the |
10fb90aa | 1130 | architecture that you are running on without resorting to loading |
2c044526 | 1131 | all of C<%Config> you can examine the content of the C<@INC> array |
10fb90aa | 1132 | like so: |
495c5fdc | 1133 | |
7d4dfb6d | 1134 | if ($^O =~ /vos/) { |
495c5fdc PG |
1135 | print "I'm on a Stratus box!\n"; |
1136 | } else { | |
1137 | print "I'm not on a Stratus box!\n"; | |
1138 | die; | |
1139 | } | |
1140 | ||
495c5fdc PG |
1141 | Also see: |
1142 | ||
1143 | =over 4 | |
1144 | ||
c997b287 | 1145 | =item * |
495c5fdc | 1146 | |
cc07ed0b | 1147 | F<README.vos> (installed as L<perlvos>) |
c997b287 GS |
1148 | |
1149 | =item * | |
1150 | ||
1151 | The VOS mailing list. | |
495c5fdc | 1152 | |
7d4dfb6d PG |
1153 | There is no specific mailing list for Perl on VOS. You can contact |
1154 | the Stratus Technologies Customer Assistance Center (CAC) for your | |
1155 | region, or you can use the contact information located in the | |
1156 | distribution files on the Stratus Anonymous FTP site. | |
495c5fdc | 1157 | |
c997b287 GS |
1158 | =item * |
1159 | ||
7d4dfb6d PG |
1160 | Stratus Technologies on the web at L<http://www.stratus.com> |
1161 | ||
1162 | =item * | |
1163 | ||
1164 | VOS Open-Source Software on the web at L<http://ftp.stratus.com/pub/vos/vos.html> | |
495c5fdc PG |
1165 | |
1166 | =back | |
1167 | ||
e41182b5 GS |
1168 | =head2 EBCDIC Platforms |
1169 | ||
6eb53dea KW |
1170 | v5.22 core Perl runs on z/OS (formerly OS/390). Theoretically it could |
1171 | run on the successors of OS/400 on AS/400 minicomputers as well as | |
1172 | VM/ESA, and BS2000 for S/390 Mainframes. Such computers use EBCDIC | |
1173 | character sets internally (usually | |
0cc436d0 | 1174 | Character Code Set ID 0037 for OS/400 and either 1047 or POSIX-BC for S/390 |
6eb53dea KW |
1175 | systems). |
1176 | ||
1177 | The rest of this section may need updating, but we don't know what it | |
1178 | should say. Please email comments to | |
1179 | L<perlbug@perl.org|mailto:perlbug@perl.org>. | |
1180 | ||
1181 | On the mainframe Perl currently works under the "Unix system | |
0cc436d0 | 1182 | services for OS/390" (formerly known as OpenEdition), VM/ESA OpenEdition, or |
2c044526 | 1183 | the BS200 POSIX-BC system (BS2000 is supported in Perl 5.6 and greater). |
522b859a | 1184 | See L<perlos390> for details. Note that for OS/400 there is also a port of |
1bcbdd38 | 1185 | Perl 5.8.1/5.10.0 or later to the PASE which is ASCII-based (as opposed to |
2890cc8c | 1186 | ILE which is EBCDIC-based), see L<perlos400>. |
e41182b5 | 1187 | |
7c5ffed3 JH |
1188 | As of R2.5 of USS for OS/390 and Version 2.3 of VM/ESA these Unix |
1189 | sub-systems do not support the C<#!> shebang trick for script invocation. | |
2c044526 | 1190 | Hence, on OS/390 and VM/ESA Perl scripts can be executed with a header |
7c5ffed3 | 1191 | similar to the following simple script: |
e41182b5 GS |
1192 | |
1193 | : # use perl | |
1194 | eval 'exec /usr/local/bin/perl -S $0 ${1+"$@"}' | |
1195 | if 0; | |
1196 | #!/usr/local/bin/perl # just a comment really | |
1197 | ||
1198 | print "Hello from perl!\n"; | |
1199 | ||
d1e3b762 GS |
1200 | OS/390 will support the C<#!> shebang trick in release 2.8 and beyond. |
1201 | Calls to C<system> and backticks can use POSIX shell syntax on all | |
1202 | S/390 systems. | |
1203 | ||
b7df3edc | 1204 | On the AS/400, if PERL5 is in your library list, you may need |
2c044526 | 1205 | to wrap your Perl scripts in a CL procedure to invoke them like so: |
6ab3f9cb GS |
1206 | |
1207 | BEGIN | |
1208 | CALL PGM(PERL5/PERL) PARM('/QOpenSys/hello.pl') | |
1209 | ENDPGM | |
1210 | ||
2c044526 | 1211 | This will invoke the Perl script F<hello.pl> in the root of the |
6ab3f9cb GS |
1212 | QOpenSys file system. On the AS/400 calls to C<system> or backticks |
1213 | must use CL syntax. | |
1214 | ||
e41182b5 | 1215 | On these platforms, bear in mind that the EBCDIC character set may have |
2c044526 | 1216 | an effect on what happens with some Perl functions (such as C<chr>, |
0a47030a GS |
1217 | C<pack>, C<print>, C<printf>, C<ord>, C<sort>, C<sprintf>, C<unpack>), as |
1218 | well as bit-fiddling with ASCII constants using operators like C<^>, C<&> | |
1219 | and C<|>, not to mention dealing with socket interfaces to ASCII computers | |
6ab3f9cb | 1220 | (see L<"Newlines">). |
e41182b5 | 1221 | |
b7df3edc GS |
1222 | Fortunately, most web servers for the mainframe will correctly |
1223 | translate the C<\n> in the following statement to its ASCII equivalent | |
2c044526 | 1224 | (C<\r> is the same under both Unix and z/OS): |
e41182b5 GS |
1225 | |
1226 | print "Content-type: text/html\r\n\r\n"; | |
1227 | ||
d1e3b762 | 1228 | The values of C<$^O> on some of these platforms includes: |
e41182b5 | 1229 | |
d1e3b762 GS |
1230 | uname $^O $Config{'archname'} |
1231 | -------------------------------------------- | |
1232 | OS/390 os390 os390 | |
1233 | OS400 os400 os400 | |
1234 | POSIX-BC posix-bc BS2000-posix-bc | |
3c075c7d | 1235 | |
e41182b5 GS |
1236 | Some simple tricks for determining if you are running on an EBCDIC |
1237 | platform could include any of the following (perhaps all): | |
1238 | ||
ce7b6f06 | 1239 | if ("\t" eq "\005") { print "EBCDIC may be spoken here!\n"; } |
e41182b5 GS |
1240 | |
1241 | if (ord('A') == 193) { print "EBCDIC may be spoken here!\n"; } | |
1242 | ||
1243 | if (chr(169) eq 'z') { print "EBCDIC may be spoken here!\n"; } | |
1244 | ||
b7df3edc | 1245 | One thing you may not want to rely on is the EBCDIC encoding |
0a47030a GS |
1246 | of punctuation characters since these may differ from code page to code |
1247 | page (and once your module or script is rumoured to work with EBCDIC, | |
1248 | folks will want it to work with all EBCDIC character sets). | |
e41182b5 GS |
1249 | |
1250 | Also see: | |
1251 | ||
1252 | =over 4 | |
1253 | ||
c997b287 GS |
1254 | =item * |
1255 | ||
6eb53dea | 1256 | L<perlos390>, L<perlos400>, L<perlbs2000>, L<perlebcdic>. |
c997b287 GS |
1257 | |
1258 | =item * | |
e41182b5 GS |
1259 | |
1260 | The perl-mvs@perl.org list is for discussion of porting issues as well as | |
1261 | general usage issues for all EBCDIC Perls. Send a message body of | |
1262 | "subscribe perl-mvs" to majordomo@perl.org. | |
1263 | ||
7ee27b7c | 1264 | =item * |
c997b287 GS |
1265 | |
1266 | AS/400 Perl information at | |
500f1b69 | 1267 | L<http://as400.rochester.ibm.com/> |
d1e3b762 | 1268 | as well as on CPAN in the F<ports/> directory. |
e41182b5 GS |
1269 | |
1270 | =back | |
1271 | ||
b8099c3d CN |
1272 | =head2 Acorn RISC OS |
1273 | ||
b7df3edc | 1274 | Because Acorns use ASCII with newlines (C<\n>) in text files as C<\012> like |
2890cc8c | 1275 | Unix, and because Unix filename emulation is turned on by default, |
b7df3edc | 1276 | most simple scripts will probably work "out of the box". The native |
6ab3f9cb | 1277 | filesystem is modular, and individual filesystems are free to be |
0a47030a | 1278 | case-sensitive or insensitive, and are usually case-preserving. Some |
b7df3edc | 1279 | native filesystems have name length limits, which file and directory |
6ab3f9cb GS |
1280 | names are silently truncated to fit. Scripts should be aware that the |
1281 | standard filesystem currently has a name length limit of B<10> | |
1282 | characters, with up to 77 items in a directory, but other filesystems | |
0a47030a | 1283 | may not impose such limitations. |
b8099c3d CN |
1284 | |
1285 | Native filenames are of the form | |
1286 | ||
6ab3f9cb | 1287 | Filesystem#Special_Field::DiskName.$.Directory.Directory.File |
dd9f0070 | 1288 | |
b8099c3d CN |
1289 | where |
1290 | ||
1291 | Special_Field is not usually present, but may contain . and $ . | |
1292 | Filesystem =~ m|[A-Za-z0-9_]| | |
1293 | DsicName =~ m|[A-Za-z0-9_/]| | |
1294 | $ represents the root directory | |
1295 | . is the path separator | |
1296 | @ is the current directory (per filesystem but machine global) | |
1297 | ^ is the parent directory | |
1298 | Directory and File =~ m|[^\0- "\.\$\%\&:\@\\^\|\177]+| | |
1299 | ||
1300 | The default filename translation is roughly C<tr|/.|./|;> | |
1301 | ||
6ab3f9cb | 1302 | Note that C<"ADFS::HardDisk.$.File" ne 'ADFS::HardDisk.$.File'> and that |
0a47030a GS |
1303 | the second stage of C<$> interpolation in regular expressions will fall |
1304 | foul of the C<$.> if scripts are not careful. | |
1305 | ||
1306 | Logical paths specified by system variables containing comma-separated | |
b7df3edc | 1307 | search lists are also allowed; hence C<System:Modules> is a valid |
0a47030a | 1308 | filename, and the filesystem will prefix C<Modules> with each section of |
6ab3f9cb | 1309 | C<System$Path> until a name is made that points to an object on disk. |
b7df3edc | 1310 | Writing to a new file C<System:Modules> would be allowed only if |
0a47030a GS |
1311 | C<System$Path> contains a single item list. The filesystem will also |
1312 | expand system variables in filenames if enclosed in angle brackets, so | |
c47ff5f1 | 1313 | C<< <System$Dir>.Modules >> would look for the file |
0a47030a | 1314 | S<C<$ENV{'System$Dir'} . 'Modules'>>. The obvious implication of this is |
c47ff5f1 | 1315 | that B<fully qualified filenames can start with C<< <> >>> and should |
0a47030a | 1316 | be protected when C<open> is used for input. |
b8099c3d CN |
1317 | |
1318 | Because C<.> was in use as a directory separator and filenames could not | |
1319 | be assumed to be unique after 10 characters, Acorn implemented the C | |
1320 | compiler to strip the trailing C<.c> C<.h> C<.s> and C<.o> suffix from | |
1321 | filenames specified in source code and store the respective files in | |
b7df3edc | 1322 | subdirectories named after the suffix. Hence files are translated: |
b8099c3d CN |
1323 | |
1324 | foo.h h.foo | |
1325 | C:foo.h C:h.foo (logical path variable) | |
1326 | sys/os.h sys.h.os (C compiler groks Unix-speak) | |
1327 | 10charname.c c.10charname | |
1328 | 10charname.o o.10charname | |
1329 | 11charname_.c c.11charname (assuming filesystem truncates at 10) | |
1330 | ||
1331 | The Unix emulation library's translation of filenames to native assumes | |
b7df3edc GS |
1332 | that this sort of translation is required, and it allows a user-defined list |
1333 | of known suffixes that it will transpose in this fashion. This may | |
92863ac5 KW |
1334 | seem transparent, but consider that with these rules F<foo/bar/baz.h> |
1335 | and F<foo/bar/h/baz> both map to F<foo.bar.h.baz>, and that C<readdir> and | |
0a47030a | 1336 | C<glob> cannot and do not attempt to emulate the reverse mapping. Other |
6ab3f9cb | 1337 | C<.>'s in filenames are translated to C</>. |
0a47030a | 1338 | |
b7df3edc | 1339 | As implied above, the environment accessed through C<%ENV> is global, and |
0a47030a | 1340 | the convention is that program specific environment variables are of the |
6ab3f9cb GS |
1341 | form C<Program$Name>. Each filesystem maintains a current directory, |
1342 | and the current filesystem's current directory is the B<global> current | |
b7df3edc GS |
1343 | directory. Consequently, sociable programs don't change the current |
1344 | directory but rely on full pathnames, and programs (and Makefiles) cannot | |
0a47030a GS |
1345 | assume that they can spawn a child process which can change the current |
1346 | directory without affecting its parent (and everyone else for that | |
1347 | matter). | |
1348 | ||
2890cc8c | 1349 | Because native operating system filehandles are global and are currently |
b7df3edc | 1350 | allocated down from 255, with 0 being a reserved value, the Unix emulation |
0a47030a GS |
1351 | library emulates Unix filehandles. Consequently, you can't rely on |
1352 | passing C<STDIN>, C<STDOUT>, or C<STDERR> to your children. | |
1353 | ||
1354 | The desire of users to express filenames of the form | |
c47ff5f1 | 1355 | C<< <Foo$Dir>.Bar >> on the command line unquoted causes problems, |
0a47030a | 1356 | too: C<``> command output capture has to perform a guessing game. It |
c47ff5f1 | 1357 | assumes that a string C<< <[^<>]+\$[^<>]> >> is a |
0a47030a | 1358 | reference to an environment variable, whereas anything else involving |
c47ff5f1 | 1359 | C<< < >> or C<< > >> is redirection, and generally manages to be 99% |
0a47030a GS |
1360 | right. Of course, the problem remains that scripts cannot rely on any |
1361 | Unix tools being available, or that any tools found have Unix-like command | |
1362 | line arguments. | |
1363 | ||
b7df3edc GS |
1364 | Extensions and XS are, in theory, buildable by anyone using free |
1365 | tools. In practice, many don't, as users of the Acorn platform are | |
1366 | used to binary distributions. MakeMaker does run, but no available | |
1367 | make currently copes with MakeMaker's makefiles; even if and when | |
1368 | this should be fixed, the lack of a Unix-like shell will cause | |
1369 | problems with makefile rules, especially lines of the form C<cd | |
1370 | sdbm && make all>, and anything using quoting. | |
b8099c3d CN |
1371 | |
1372 | "S<RISC OS>" is the proper name for the operating system, but the value | |
1373 | in C<$^O> is "riscos" (because we don't like shouting). | |
1374 | ||
e41182b5 GS |
1375 | =head2 Other perls |
1376 | ||
b7df3edc | 1377 | Perl has been ported to many platforms that do not fit into any of |
b6c36746 | 1378 | the categories listed above. Some, such as AmigaOS, |
cd86ed9d JV |
1379 | QNX, Plan 9, and VOS, have been well-integrated into the standard |
1380 | Perl source code kit. You may need to see the F<ports/> directory | |
1381 | on CPAN for information, and possibly binaries, for the likes of: | |
1382 | aos, Atari ST, lynxos, riscos, Novell Netware, Tandem Guardian, | |
1383 | I<etc.> (Yes, we know that some of these OSes may fall under the | |
1384 | Unix category, but we are not a standards body.) | |
e41182b5 | 1385 | |
d1e3b762 GS |
1386 | Some approximate operating system names and their C<$^O> values |
1387 | in the "OTHER" category include: | |
1388 | ||
1389 | OS $^O $Config{'archname'} | |
1390 | ------------------------------------------ | |
1391 | Amiga DOS amigaos m68k-amigos | |
d1e3b762 | 1392 | |
e41182b5 GS |
1393 | See also: |
1394 | ||
1395 | =over 4 | |
1396 | ||
c997b287 GS |
1397 | =item * |
1398 | ||
1399 | Amiga, F<README.amiga> (installed as L<perlamiga>). | |
1400 | ||
1401 | =item * | |
d1e3b762 | 1402 | |
6ab3f9cb | 1403 | A free perl5-based PERL.NLM for Novell Netware is available in |
500f1b69 | 1404 | precompiled binary and source code form from L<http://www.novell.com/> |
6ab3f9cb | 1405 | as well as from CPAN. |
e41182b5 | 1406 | |
13a2d996 | 1407 | =item * |
c997b287 | 1408 | |
e6f03d26 | 1409 | S<Plan 9>, F<README.plan9> |
d1e3b762 | 1410 | |
e41182b5 GS |
1411 | =back |
1412 | ||
e41182b5 GS |
1413 | =head1 FUNCTION IMPLEMENTATIONS |
1414 | ||
b7df3edc GS |
1415 | Listed below are functions that are either completely unimplemented |
1416 | or else have been implemented differently on various platforms. | |
1417 | Following each description will be, in parentheses, a list of | |
1418 | platforms that the description applies to. | |
e41182b5 | 1419 | |
b7df3edc GS |
1420 | The list may well be incomplete, or even wrong in some places. When |
1421 | in doubt, consult the platform-specific README files in the Perl | |
1422 | source distribution, and any other documentation resources accompanying | |
1423 | a given port. | |
e41182b5 | 1424 | |
0a47030a | 1425 | Be aware, moreover, that even among Unix-ish systems there are variations. |
e41182b5 | 1426 | |
b7df3edc | 1427 | For many functions, you can also query C<%Config>, exported by |
2c044526 | 1428 | default from the C<Config> module. For example, to check whether the |
b7df3edc GS |
1429 | platform has the C<lstat> call, check C<$Config{d_lstat}>. See |
1430 | L<Config> for a full description of available variables. | |
e41182b5 GS |
1431 | |
1432 | =head2 Alphabetical Listing of Perl Functions | |
1433 | ||
1434 | =over 8 | |
1435 | ||
e41182b5 GS |
1436 | =item -X |
1437 | ||
038ae9a4 SH |
1438 | C<-w> only inspects the read-only file attribute (FILE_ATTRIBUTE_READONLY), |
1439 | which determines whether the directory can be deleted, not whether it can | |
1440 | be written to. Directories always have read and write access unless denied | |
1441 | by discretionary access control lists (DACLs). (S<Win32>) | |
1442 | ||
b7df3edc GS |
1443 | C<-r>, C<-w>, C<-x>, and C<-o> tell whether the file is accessible, |
1444 | which may not reflect UIC-based file protections. (VMS) | |
e41182b5 | 1445 | |
b8099c3d CN |
1446 | C<-s> by name on an open file will return the space reserved on disk, |
1447 | rather than the current extent. C<-s> on an open filehandle returns the | |
b7df3edc | 1448 | current size. (S<RISC OS>) |
b8099c3d | 1449 | |
e41182b5 | 1450 | C<-R>, C<-W>, C<-X>, C<-O> are indistinguishable from C<-r>, C<-w>, |
204ad8d5 | 1451 | C<-x>, C<-o>. (Win32, VMS, S<RISC OS>) |
e41182b5 | 1452 | |
287a962e | 1453 | C<-g>, C<-k>, C<-l>, C<-u>, C<-A> are not particularly meaningful. |
b8099c3d | 1454 | (Win32, VMS, S<RISC OS>) |
e41182b5 | 1455 | |
287a962e JD |
1456 | C<-p> is not particularly meaningful. (VMS, S<RISC OS>) |
1457 | ||
e41182b5 GS |
1458 | C<-d> is true if passed a device spec without an explicit directory. |
1459 | (VMS) | |
1460 | ||
e41182b5 | 1461 | C<-x> (or C<-X>) determine if a file ends in one of the executable |
b7df3edc | 1462 | suffixes. C<-S> is meaningless. (Win32) |
e41182b5 | 1463 | |
b8099c3d CN |
1464 | C<-x> (or C<-X>) determine if a file has an executable file type. |
1465 | (S<RISC OS>) | |
1466 | ||
aca72608 JD |
1467 | =item alarm |
1468 | ||
1469 | Emulated using timers that must be explicitly polled whenever Perl | |
1470 | wants to dispatch "safe signals" and therefore cannot interrupt | |
1471 | blocking system calls. (Win32) | |
1472 | ||
47cd99a4 | 1473 | =item atan2 |
519bc777 RGS |
1474 | |
1475 | Due to issues with various CPUs, math libraries, compilers, and standards, | |
1476 | results for C<atan2()> may vary depending on any combination of the above. | |
1477 | Perl attempts to conform to the Open Group/IEEE standards for the results | |
1478 | returned from C<atan2()>, but cannot force the issue if the system Perl is | |
2890cc8c | 1479 | run on does not allow it. (Tru64, HP-UX 10.20) |
519bc777 | 1480 | |
2890cc8c | 1481 | The current version of the standards for C<atan2()> is available at |
519bc777 RGS |
1482 | L<http://www.opengroup.org/onlinepubs/009695399/functions/atan2.html>. |
1483 | ||
47cd99a4 | 1484 | =item binmode |
e41182b5 | 1485 | |
204ad8d5 | 1486 | Meaningless. (S<RISC OS>) |
e41182b5 GS |
1487 | |
1488 | Reopens file and restores pointer; if function fails, underlying | |
1489 | filehandle may be closed, or pointer may be in a different position. | |
1490 | (VMS) | |
1491 | ||
1492 | The value returned by C<tell> may be affected after the call, and | |
1493 | the filehandle may be flushed. (Win32) | |
1494 | ||
47cd99a4 | 1495 | =item chmod |
e41182b5 | 1496 | |
e41182b5 GS |
1497 | Only good for changing "owner" read-write access, "group", and "other" |
1498 | bits are meaningless. (Win32) | |
1499 | ||
b8099c3d CN |
1500 | Only good for changing "owner" and "other" read-write access. (S<RISC OS>) |
1501 | ||
495c5fdc PG |
1502 | Access permissions are mapped onto VOS access-control list changes. (VOS) |
1503 | ||
4e51f8e4 | 1504 | The actual permissions set depend on the value of the C<CYGWIN> |
789f0d36 | 1505 | in the SYSTEM environment settings. (Cygwin) |
4e51f8e4 | 1506 | |
2c044526 | 1507 | Setting the exec bit on some locations (generally F</sdcard>) will return true |
43b08d74 BF |
1508 | but not actually set the bit. (Android) |
1509 | ||
47cd99a4 | 1510 | =item chown |
e41182b5 | 1511 | |
204ad8d5 | 1512 | Not implemented. (Win32, S<Plan 9>, S<RISC OS>) |
e41182b5 GS |
1513 | |
1514 | Does nothing, but won't fail. (Win32) | |
1515 | ||
3fd80bd6 PG |
1516 | A little funky, because VOS's notion of ownership is a little funky (VOS). |
1517 | ||
e41182b5 GS |
1518 | =item chroot |
1519 | ||
043fec90 | 1520 | Not implemented. (Win32, VMS, S<Plan 9>, S<RISC OS>, VOS) |
e41182b5 | 1521 | |
47cd99a4 | 1522 | =item crypt |
e41182b5 GS |
1523 | |
1524 | May not be available if library or source was not provided when building | |
b8099c3d | 1525 | perl. (Win32) |
e41182b5 | 1526 | |
43b08d74 BF |
1527 | Not implemented. (Android) |
1528 | ||
47cd99a4 | 1529 | =item dbmclose |
e41182b5 | 1530 | |
e6f03d26 | 1531 | Not implemented. (VMS, S<Plan 9>, VOS) |
e41182b5 | 1532 | |
47cd99a4 | 1533 | =item dbmopen |
e41182b5 | 1534 | |
e6f03d26 | 1535 | Not implemented. (VMS, S<Plan 9>, VOS) |
e41182b5 | 1536 | |
47cd99a4 | 1537 | =item dump |
e41182b5 | 1538 | |
204ad8d5 | 1539 | Not useful. (S<RISC OS>) |
e41182b5 | 1540 | |
84d78eb7 | 1541 | Not supported. (Cygwin, Win32) |
e41182b5 | 1542 | |
b8099c3d | 1543 | Invokes VMS debugger. (VMS) |
e41182b5 | 1544 | |
47cd99a4 | 1545 | =item exec |
e41182b5 | 1546 | |
94d4006a | 1547 | C<exec LIST> without the use of indirect object syntax (C<exec PROGRAM LIST>) |
2c044526 | 1548 | may fall back to trying the shell if the first C<spawn()> fails. (Win32) |
94d4006a | 1549 | |
0f897271 GS |
1550 | Does not automatically flush output handles on some platforms. |
1551 | (SunOS, Solaris, HP-UX) | |
1552 | ||
af8bb25a FC |
1553 | Not supported. (Symbian OS) |
1554 | ||
fe12c0e8 MS |
1555 | =item exit |
1556 | ||
2c044526 KW |
1557 | Emulates Unix C<exit()> (which considers C<exit 1> to indicate an error) by |
1558 | mapping the C<1> to C<SS$_ABORT> (C<44>). This behavior may be overridden | |
1559 | with the pragma C<use vmsish 'exit'>. As with the CRTL's C<exit()> | |
1560 | function, C<exit 0> is also mapped to an exit status of C<SS$_NORMAL> | |
1561 | (C<1>); this mapping cannot be overridden. Any other argument to | |
1562 | C<exit()> | |
016930a6 JM |
1563 | is used directly as Perl's exit status. On VMS, unless the future |
1564 | POSIX_EXIT mode is enabled, the exit code should always be a valid | |
1565 | VMS exit code and not a generic number. When the POSIX_EXIT mode is | |
1566 | enabled, a generic number will be encoded in a method compatible with | |
1567 | the C library _POSIX_EXIT macro so that it can be decoded by other | |
1568 | programs, particularly ones written in C, like the GNV package. (VMS) | |
fe12c0e8 | 1569 | |
2890cc8c CBW |
1570 | C<exit()> resets file pointers, which is a problem when called |
1571 | from a child process (created by C<fork()>) in C<BEGIN>. | |
bef2c191 RB |
1572 | A workaround is to use C<POSIX::_exit>. (Solaris) |
1573 | ||
1574 | exit unless $Config{archname} =~ /\bsolaris\b/; | |
1575 | require POSIX and POSIX::_exit(0); | |
1576 | ||
47cd99a4 | 1577 | =item fcntl |
e41182b5 | 1578 | |
016930a6 | 1579 | Not implemented. (Win32) |
6a065175 | 1580 | |
016930a6 | 1581 | Some functions available based on the version of VMS. (VMS) |
e41182b5 | 1582 | |
47cd99a4 | 1583 | =item flock |
e41182b5 | 1584 | |
204ad8d5 | 1585 | Not implemented (VMS, S<RISC OS>, VOS). |
e41182b5 | 1586 | |
e41182b5 GS |
1587 | =item fork |
1588 | ||
043fec90 | 1589 | Not implemented. (AmigaOS, S<RISC OS>, VMS) |
0f897271 GS |
1590 | |
1591 | Emulated using multiple interpreters. See L<perlfork>. (Win32) | |
1592 | ||
1593 | Does not automatically flush output handles on some platforms. | |
1594 | (SunOS, Solaris, HP-UX) | |
e41182b5 GS |
1595 | |
1596 | =item getlogin | |
1597 | ||
204ad8d5 | 1598 | Not implemented. (S<RISC OS>) |
e41182b5 | 1599 | |
47cd99a4 | 1600 | =item getpgrp |
e41182b5 | 1601 | |
204ad8d5 | 1602 | Not implemented. (Win32, VMS, S<RISC OS>) |
e41182b5 GS |
1603 | |
1604 | =item getppid | |
1605 | ||
204ad8d5 | 1606 | Not implemented. (Win32, S<RISC OS>) |
e41182b5 | 1607 | |
47cd99a4 | 1608 | =item getpriority |
e41182b5 | 1609 | |
043fec90 | 1610 | Not implemented. (Win32, VMS, S<RISC OS>, VOS) |
e41182b5 | 1611 | |
47cd99a4 | 1612 | =item getpwnam |
e41182b5 | 1613 | |
204ad8d5 | 1614 | Not implemented. (Win32) |
e41182b5 | 1615 | |
b8099c3d CN |
1616 | Not useful. (S<RISC OS>) |
1617 | ||
47cd99a4 | 1618 | =item getgrnam |
e41182b5 | 1619 | |
204ad8d5 | 1620 | Not implemented. (Win32, VMS, S<RISC OS>) |
e41182b5 | 1621 | |
47cd99a4 | 1622 | =item getnetbyname |
e41182b5 | 1623 | |
43b08d74 | 1624 | Not implemented. (Android, Win32, S<Plan 9>) |
e41182b5 | 1625 | |
47cd99a4 | 1626 | =item getpwuid |
e41182b5 | 1627 | |
204ad8d5 | 1628 | Not implemented. (Win32) |
e41182b5 | 1629 | |
b8099c3d CN |
1630 | Not useful. (S<RISC OS>) |
1631 | ||
47cd99a4 | 1632 | =item getgrgid |
e41182b5 | 1633 | |
204ad8d5 | 1634 | Not implemented. (Win32, VMS, S<RISC OS>) |
e41182b5 | 1635 | |
47cd99a4 | 1636 | =item getnetbyaddr |
e41182b5 | 1637 | |
43b08d74 | 1638 | Not implemented. (Android, Win32, S<Plan 9>) |
e41182b5 | 1639 | |
47cd99a4 | 1640 | =item getprotobynumber |
e41182b5 | 1641 | |
43b08d74 BF |
1642 | Not implemented. (Android) |
1643 | ||
47cd99a4 | 1644 | =item getservbyport |
e41182b5 | 1645 | |
e41182b5 GS |
1646 | =item getpwent |
1647 | ||
43b08d74 | 1648 | Not implemented. (Android, Win32) |
e41182b5 GS |
1649 | |
1650 | =item getgrent | |
1651 | ||
43b08d74 | 1652 | Not implemented. (Android, Win32, VMS) |
e41182b5 | 1653 | |
ef5a6dd7 JH |
1654 | =item gethostbyname |
1655 | ||
1656 | C<gethostbyname('localhost')> does not work everywhere: you may have | |
204ad8d5 | 1657 | to use C<gethostbyname('127.0.0.1')>. (S<Irix 5>) |
ef5a6dd7 | 1658 | |
e41182b5 GS |
1659 | =item gethostent |
1660 | ||
204ad8d5 | 1661 | Not implemented. (Win32) |
e41182b5 GS |
1662 | |
1663 | =item getnetent | |
1664 | ||
43b08d74 | 1665 | Not implemented. (Android, Win32, S<Plan 9>) |
e41182b5 GS |
1666 | |
1667 | =item getprotoent | |
1668 | ||
43b08d74 | 1669 | Not implemented. (Android, Win32, S<Plan 9>) |
e41182b5 GS |
1670 | |
1671 | =item getservent | |
1672 | ||
e6f03d26 | 1673 | Not implemented. (Win32, S<Plan 9>) |
e41182b5 | 1674 | |
43b08d74 BF |
1675 | =item seekdir |
1676 | ||
1677 | Not implemented. (Android) | |
1678 | ||
47cd99a4 | 1679 | =item sethostent |
e41182b5 | 1680 | |
43b08d74 | 1681 | Not implemented. (Android, Win32, S<Plan 9>, S<RISC OS>) |
e41182b5 | 1682 | |
47cd99a4 | 1683 | =item setnetent |
e41182b5 | 1684 | |
204ad8d5 | 1685 | Not implemented. (Win32, S<Plan 9>, S<RISC OS>) |
e41182b5 | 1686 | |
47cd99a4 | 1687 | =item setprotoent |
e41182b5 | 1688 | |
43b08d74 | 1689 | Not implemented. (Android, Win32, S<Plan 9>, S<RISC OS>) |
e41182b5 | 1690 | |
47cd99a4 | 1691 | =item setservent |
e41182b5 | 1692 | |
e6f03d26 | 1693 | Not implemented. (S<Plan 9>, Win32, S<RISC OS>) |
e41182b5 GS |
1694 | |
1695 | =item endpwent | |
1696 | ||
b5afd346 | 1697 | Not implemented. (Win32) |
e41182b5 | 1698 | |
43b08d74 BF |
1699 | Either not implemented or a no-op. (Android) |
1700 | ||
e41182b5 GS |
1701 | =item endgrent |
1702 | ||
43b08d74 | 1703 | Not implemented. (Android, S<RISC OS>, VMS, Win32) |
e41182b5 GS |
1704 | |
1705 | =item endhostent | |
1706 | ||
43b08d74 | 1707 | Not implemented. (Android, Win32) |
e41182b5 GS |
1708 | |
1709 | =item endnetent | |
1710 | ||
43b08d74 | 1711 | Not implemented. (Android, Win32, S<Plan 9>) |
e41182b5 GS |
1712 | |
1713 | =item endprotoent | |
1714 | ||
43b08d74 | 1715 | Not implemented. (Android, Win32, S<Plan 9>) |
e41182b5 GS |
1716 | |
1717 | =item endservent | |
1718 | ||
e6f03d26 | 1719 | Not implemented. (S<Plan 9>, Win32) |
e41182b5 GS |
1720 | |
1721 | =item getsockopt SOCKET,LEVEL,OPTNAME | |
1722 | ||
e6f03d26 | 1723 | Not implemented. (S<Plan 9>) |
e41182b5 | 1724 | |
e41182b5 GS |
1725 | =item glob |
1726 | ||
2c044526 | 1727 | This operator is implemented via the C<File::Glob> extension on most |
63f87e49 | 1728 | platforms. See L<File::Glob> for portability information. |
b8099c3d | 1729 | |
62aa5637 MS |
1730 | =item gmtime |
1731 | ||
2c044526 | 1732 | In theory, C<gmtime()> is reliable from -2**63 to 2**63-1. However, |
461d5a49 MS |
1733 | because work arounds in the implementation use floating point numbers, |
1734 | it will become inaccurate as the time gets larger. This is a bug and | |
1735 | will be fixed in the future. | |
62aa5637 | 1736 | |
10fb90aa PG |
1737 | On VOS, time values are 32-bit quantities. |
1738 | ||
e41182b5 GS |
1739 | =item ioctl FILEHANDLE,FUNCTION,SCALAR |
1740 | ||
1741 | Not implemented. (VMS) | |
1742 | ||
2c044526 | 1743 | Available only for socket handles, and it does what the C<ioctlsocket()> call |
e41182b5 GS |
1744 | in the Winsock API does. (Win32) |
1745 | ||
b8099c3d CN |
1746 | Available only for socket handles. (S<RISC OS>) |
1747 | ||
47cd99a4 | 1748 | =item kill |
e41182b5 | 1749 | |
862b5365 | 1750 | Not implemented, hence not useful for taint checking. (S<RISC OS>) |
e41182b5 | 1751 | |
63f87e49 GS |
1752 | C<kill()> doesn't have the semantics of C<raise()>, i.e. it doesn't send |
1753 | a signal to the identified process like it does on Unix platforms. | |
2c044526 | 1754 | Instead C<kill($sig, $pid)> terminates the process identified by C<$pid>, |
63f87e49 GS |
1755 | and makes it exit immediately with exit status $sig. As in Unix, if |
1756 | $sig is 0 and the specified process exists, it returns true without | |
1757 | actually terminating it. (Win32) | |
e41182b5 | 1758 | |
2c044526 | 1759 | C<kill(-9, $pid)> will terminate the process specified by C<$pid> and |
d0302514 JD |
1760 | recursively all child processes owned by it. This is different from |
1761 | the Unix semantics, where the signal will be delivered to all | |
1762 | processes in the same process group as the process specified by | |
1763 | $pid. (Win32) | |
1764 | ||
96f902ff CB |
1765 | A pid of -1 indicating all processes on the system is not currently |
1766 | supported. (VMS) | |
016930a6 | 1767 | |
47cd99a4 | 1768 | =item link |
e41182b5 | 1769 | |
b5afd346 | 1770 | Not implemented. (S<RISC OS>, VOS) |
e41182b5 | 1771 | |
433acd8a JH |
1772 | Link count not updated because hard links are not quite that hard |
1773 | (They are sort of half-way between hard and soft links). (AmigaOS) | |
1774 | ||
63d6c08b JD |
1775 | Hard links are implemented on Win32 under NTFS only. They are |
1776 | natively supported on Windows 2000 and later. On Windows NT they | |
1777 | are implemented using the Windows POSIX subsystem support and the | |
1778 | Perl process will need Administrator or Backup Operator privileges | |
1779 | to create hard links. | |
a3dfe201 | 1780 | |
016930a6 JM |
1781 | Available on 64 bit OpenVMS 8.2 and later. (VMS) |
1782 | ||
62aa5637 MS |
1783 | =item localtime |
1784 | ||
a61fc69c | 1785 | localtime() has the same range as L</gmtime>, but because time zone |
dc164757 MS |
1786 | rules change its accuracy for historical and future times may degrade |
1787 | but usually by no more than an hour. | |
62aa5637 | 1788 | |
e41182b5 GS |
1789 | =item lstat |
1790 | ||
016930a6 | 1791 | Not implemented. (S<RISC OS>) |
e41182b5 | 1792 | |
63f87e49 | 1793 | Return values (especially for device and inode) may be bogus. (Win32) |
e41182b5 | 1794 | |
47cd99a4 | 1795 | =item msgctl |
e41182b5 | 1796 | |
47cd99a4 | 1797 | =item msgget |
e41182b5 | 1798 | |
47cd99a4 | 1799 | =item msgsnd |
e41182b5 | 1800 | |
47cd99a4 | 1801 | =item msgrcv |
e41182b5 | 1802 | |
43b08d74 | 1803 | Not implemented. (Android, Win32, VMS, S<Plan 9>, S<RISC OS>, VOS) |
e41182b5 | 1804 | |
47cd99a4 | 1805 | =item open |
e41182b5 | 1806 | |
204ad8d5 | 1807 | open to C<|-> and C<-|> are unsupported. (Win32, S<RISC OS>) |
e41182b5 | 1808 | |
0f897271 GS |
1809 | Opening a process does not automatically flush output handles on some |
1810 | platforms. (SunOS, Solaris, HP-UX) | |
1811 | ||
e41182b5 GS |
1812 | =item readlink |
1813 | ||
b8099c3d | 1814 | Not implemented. (Win32, VMS, S<RISC OS>) |
e41182b5 | 1815 | |
47cd99a4 | 1816 | =item rename |
c9b2b9d4 SS |
1817 | |
1818 | Can't move directories between directories on different logical volumes. (Win32) | |
1819 | ||
3ba4b5c1 JD |
1820 | =item rewinddir |
1821 | ||
2c044526 KW |
1822 | Will not cause C<readdir()> to re-read the directory stream. The entries |
1823 | already read before the C<rewinddir()> call will just be returned again | |
3ba4b5c1 JD |
1824 | from a cache buffer. (Win32) |
1825 | ||
47cd99a4 | 1826 | =item select |
e41182b5 | 1827 | |
689c5c24 | 1828 | Only implemented on sockets. (Win32, VMS) |
e41182b5 | 1829 | |
b8099c3d CN |
1830 | Only reliable on sockets. (S<RISC OS>) |
1831 | ||
76e05f0b | 1832 | Note that the C<select FILEHANDLE> form is generally portable. |
63f87e49 | 1833 | |
47cd99a4 | 1834 | =item semctl |
e41182b5 | 1835 | |
47cd99a4 | 1836 | =item semget |
e41182b5 | 1837 | |
47cd99a4 | 1838 | =item semop |
e41182b5 | 1839 | |
43b08d74 | 1840 | Not implemented. (Android, Win32, VMS, S<RISC OS>) |
e41182b5 | 1841 | |
a3dfe201 GS |
1842 | =item setgrent |
1843 | ||
43b08d74 | 1844 | Not implemented. (Android, VMS, Win32, S<RISC OS>) |
a3dfe201 | 1845 | |
47cd99a4 | 1846 | =item setpgrp |
e41182b5 | 1847 | |
204ad8d5 | 1848 | Not implemented. (Win32, VMS, S<RISC OS>, VOS) |
e41182b5 | 1849 | |
47cd99a4 | 1850 | =item setpriority |
e41182b5 | 1851 | |
204ad8d5 | 1852 | Not implemented. (Win32, VMS, S<RISC OS>, VOS) |
e41182b5 | 1853 | |
a3dfe201 GS |
1854 | =item setpwent |
1855 | ||
43b08d74 | 1856 | Not implemented. (Android, Win32, S<RISC OS>) |
a3dfe201 | 1857 | |
47cd99a4 | 1858 | =item setsockopt |
e41182b5 | 1859 | |
e6f03d26 | 1860 | Not implemented. (S<Plan 9>) |
e41182b5 | 1861 | |
47cd99a4 | 1862 | =item shmctl |
e41182b5 | 1863 | |
47cd99a4 | 1864 | =item shmget |
e41182b5 | 1865 | |
47cd99a4 | 1866 | =item shmread |
e41182b5 | 1867 | |
47cd99a4 | 1868 | =item shmwrite |
e41182b5 | 1869 | |
43b08d74 | 1870 | Not implemented. (Android, Win32, VMS, S<RISC OS>) |
e41182b5 | 1871 | |
001e9f89 DD |
1872 | =item sleep |
1873 | ||
3cd50447 | 1874 | Emulated using synchronization functions such that it can be |
2c044526 | 1875 | interrupted by C<alarm()>, and limited to a maximum of 4294967 seconds, |
3cd50447 | 1876 | approximately 49 days. (Win32) |
001e9f89 | 1877 | |
47cd99a4 | 1878 | =item sockatmark |
80cbd5ad JH |
1879 | |
1880 | A relatively recent addition to socket functions, may not | |
e1020413 | 1881 | be implemented even in Unix platforms. |
80cbd5ad | 1882 | |
47cd99a4 | 1883 | =item socketpair |
e41182b5 | 1884 | |
043fec90 | 1885 | Not implemented. (S<RISC OS>) |
10fb90aa | 1886 | |
016930a6 | 1887 | Available on 64 bit OpenVMS 8.2 and later. (VMS) |
e41182b5 | 1888 | |
e41182b5 GS |
1889 | =item stat |
1890 | ||
d62e1b7f JH |
1891 | Platforms that do not have rdev, blksize, or blocks will return these |
1892 | as '', so numeric comparison or manipulation of these fields may cause | |
1893 | 'not numeric' warnings. | |
1894 | ||
3f1f789b | 1895 | ctime not supported on UFS (S<Mac OS X>). |
e41182b5 | 1896 | |
95a3fe12 MS |
1897 | ctime is creation time instead of inode change time (Win32). |
1898 | ||
e41182b5 GS |
1899 | device and inode are not meaningful. (Win32) |
1900 | ||
1901 | device and inode are not necessarily reliable. (VMS) | |
1902 | ||
b8099c3d CN |
1903 | mtime, atime and ctime all return the last modification time. Device and |
1904 | inode are not necessarily reliable. (S<RISC OS>) | |
1905 | ||
d62e1b7f JH |
1906 | dev, rdev, blksize, and blocks are not available. inode is not |
1907 | meaningful and will differ between stat calls on the same file. (os2) | |
1908 | ||
2c044526 KW |
1909 | some versions of cygwin when doing a C<stat("foo")> and if not finding it |
1910 | may then attempt to C<stat("foo.exe")> (Cygwin) | |
73e9292c | 1911 | |
2c044526 | 1912 | On Win32 C<stat()> needs to open the file to determine the link count |
1fafdf34 | 1913 | and update attributes that may have been changed through hard links. |
2c044526 | 1914 | Setting C<${^WIN32_SLOPPY_STAT}> to a true value speeds up C<stat()> by |
1fafdf34 JD |
1915 | not performing this operation. (Win32) |
1916 | ||
47cd99a4 | 1917 | =item symlink |
e41182b5 | 1918 | |
c73b03b7 JM |
1919 | Not implemented. (Win32, S<RISC OS>) |
1920 | ||
1921 | Implemented on 64 bit VMS 8.3. VMS requires the symbolic link to be in Unix | |
1922 | syntax if it is intended to resolve to a valid path. | |
e41182b5 | 1923 | |
47cd99a4 | 1924 | =item syscall |
e41182b5 | 1925 | |
043fec90 | 1926 | Not implemented. (Win32, VMS, S<RISC OS>, VOS) |
e41182b5 | 1927 | |
47cd99a4 | 1928 | =item sysopen |
f34d0673 | 1929 | |
dd9f0070 | 1930 | The traditional "0", "1", and "2" MODEs are implemented with different |
322422de GS |
1931 | numeric values on some systems. The flags exported by C<Fcntl> |
1932 | (O_RDONLY, O_WRONLY, O_RDWR) should work everywhere though. (S<Mac | |
043fec90 | 1933 | OS>, OS/390) |
f34d0673 | 1934 | |
47cd99a4 | 1935 | =item system |
e41182b5 | 1936 | |
e41182b5 | 1937 | As an optimization, may not call the command shell specified in |
b7df3edc | 1938 | C<$ENV{PERL5SHELL}>. C<system(1, @args)> spawns an external |
e41182b5 GS |
1939 | process and immediately returns its process designator, without |
1940 | waiting for it to terminate. Return value may be used subsequently | |
2c044526 KW |
1941 | in C<wait> or C<waitpid>. Failure to C<spawn()> a subprocess is indicated |
1942 | by setting C<$?> to S<C<"255 << 8">>. C<$?> is set in a way compatible with | |
1943 | Unix (i.e. the exitstatus of the subprocess is obtained by S<C<"$? >> 8">>, | |
63f87e49 | 1944 | as described in the documentation). (Win32) |
e41182b5 | 1945 | |
b8099c3d CN |
1946 | There is no shell to process metacharacters, and the native standard is |
1947 | to pass a command line terminated by "\n" "\r" or "\0" to the spawned | |
c47ff5f1 | 1948 | program. Redirection such as C<< > foo >> is performed (if at all) by |
b8099c3d CN |
1949 | the run time library of the spawned program. C<system> I<list> will call |
1950 | the Unix emulation library's C<exec> emulation, which attempts to provide | |
1951 | emulation of the stdin, stdout, stderr in force in the parent, providing | |
1952 | the child program uses a compatible version of the emulation library. | |
1953 | I<scalar> will call the native command line direct and no such emulation | |
1954 | of a child Unix program will exists. Mileage B<will> vary. (S<RISC OS>) | |
1955 | ||
94d4006a | 1956 | C<system LIST> without the use of indirect object syntax (C<system PROGRAM LIST>) |
2c044526 | 1957 | may fall back to trying the shell if the first C<spawn()> fails. (Win32) |
94d4006a | 1958 | |
0f897271 GS |
1959 | Does not automatically flush output handles on some platforms. |
1960 | (SunOS, Solaris, HP-UX) | |
1961 | ||
9bc98430 CB |
1962 | The return value is POSIX-like (shifted up by 8 bits), which only allows |
1963 | room for a made-up value derived from the severity bits of the native | |
2890cc8c | 1964 | 32-bit condition code (unless overridden by C<use vmsish 'status'>). |
016930a6 JM |
1965 | If the native condition code is one that has a POSIX value encoded, the |
1966 | POSIX value will be decoded to extract the expected exit value. | |
9bc98430 CB |
1967 | For more details see L<perlvms/$?>. (VMS) |
1968 | ||
43b08d74 BF |
1969 | =item telldir |
1970 | ||
1971 | Not implemented. (Android) | |
1972 | ||
e41182b5 GS |
1973 | =item times |
1974 | ||
63f87e49 GS |
1975 | "cumulative" times will be bogus. On anything other than Windows NT |
1976 | or Windows 2000, "system" time will be bogus, and "user" time is | |
2c044526 | 1977 | actually the time returned by the C<clock()> function in the C runtime |
63f87e49 | 1978 | library. (Win32) |
e41182b5 | 1979 | |
b8099c3d CN |
1980 | Not useful. (S<RISC OS>) |
1981 | ||
47cd99a4 | 1982 | =item truncate |
e41182b5 | 1983 | |
6d738113 | 1984 | Not implemented. (Older versions of VMS) |
e41182b5 | 1985 | |
3fd80bd6 | 1986 | Truncation to same-or-shorter lengths only. (VOS) |
495c5fdc | 1987 | |
4cfdb94f | 1988 | If a FILEHANDLE is supplied, it must be writable and opened in append |
e71a7dc8 | 1989 | mode (i.e., use C<<< open(FH, '>>filename') >>> |
4cfdb94f GS |
1990 | or C<sysopen(FH,...,O_APPEND|O_RDWR)>. If a filename is supplied, it |
1991 | should not be held open elsewhere. (Win32) | |
1992 | ||
e41182b5 GS |
1993 | =item umask |
1994 | ||
1bcbdd38 | 1995 | Returns undef where unavailable. |
e41182b5 | 1996 | |
b7df3edc GS |
1997 | C<umask> works but the correct permissions are set only when the file |
1998 | is finally closed. (AmigaOS) | |
433acd8a | 1999 | |
47cd99a4 | 2000 | =item utime |
e41182b5 | 2001 | |
b6c36746 | 2002 | Only the modification time is updated. (VMS, S<RISC OS>) |
e41182b5 | 2003 | |
322422de | 2004 | May not behave as expected. Behavior depends on the C runtime |
2c044526 | 2005 | library's implementation of C<utime()>, and the filesystem being |
322422de GS |
2006 | used. The FAT filesystem typically does not support an "access |
2007 | time" field, and it may limit timestamps to a granularity of | |
2008 | two seconds. (Win32) | |
e41182b5 GS |
2009 | |
2010 | =item wait | |
2011 | ||
47cd99a4 | 2012 | =item waitpid |
e41182b5 | 2013 | |
e41182b5 | 2014 | Can only be applied to process handles returned for processes spawned |
a6f858fb | 2015 | using C<system(1, ...)> or pseudo processes created with C<fork()>. (Win32) |
e41182b5 | 2016 | |
b8099c3d CN |
2017 | Not useful. (S<RISC OS>) |
2018 | ||
e41182b5 GS |
2019 | =back |
2020 | ||
2021 | ||
7c35b6af | 2022 | =head1 Supported Platforms |
ba58ab26 | 2023 | |
7c35b6af RGS |
2024 | The following platforms are known to build Perl 5.12 (as of April 2010, |
2025 | its release date) from the standard source code distribution available | |
500f1b69 | 2026 | at L<http://www.cpan.org/src> |
bb377ba2 | 2027 | |
bb377ba2 JV |
2028 | =over |
2029 | ||
2030 | =item Linux (x86, ARM, IA64) | |
2031 | ||
e0d9a2c8 | 2032 | =item HP-UX |
bb377ba2 JV |
2033 | |
2034 | =item AIX | |
2035 | ||
2036 | =item Win32 | |
2037 | ||
2038 | =over | |
2039 | ||
2040 | =item Windows 2000 | |
2041 | ||
2042 | =item Windows XP | |
2043 | ||
2044 | =item Windows Server 2003 | |
2045 | ||
2046 | =item Windows Vista | |
2047 | ||
2048 | =item Windows Server 2008 | |
2049 | ||
3b665c47 JD |
2050 | =item Windows 7 |
2051 | ||
bb377ba2 JV |
2052 | =back |
2053 | ||
2d9ede6e JH |
2054 | =item Cygwin |
2055 | ||
67e52905 TC |
2056 | Some tests are known to fail: |
2057 | ||
2058 | =over | |
2059 | ||
2060 | =item * | |
2061 | ||
cb0ee57a | 2062 | F<ext/XS-APItest/t/call_checker.t> - see |
67e52905 TC |
2063 | L<https://rt.perl.org/Ticket/Display.html?id=78502> |
2064 | ||
2065 | =item * | |
2066 | ||
2067 | F<dist/I18N-Collate/t/I18N-Collate.t> | |
2068 | ||
2069 | =item * | |
2070 | ||
2071 | F<ext/Win32CORE/t/win32core.t> - may fail on recent cygwin installs. | |
2072 | ||
2073 | =back | |
2074 | ||
bb377ba2 JV |
2075 | =item Solaris (x86, SPARC) |
2076 | ||
1b0ab010 JV |
2077 | =item OpenVMS |
2078 | ||
2079 | =over | |
2080 | ||
2081 | =item Alpha (7.2 and later) | |
2082 | ||
2083 | =item I64 (8.2 and later) | |
2084 | ||
2085 | =back | |
bb377ba2 JV |
2086 | |
2087 | =item Symbian | |
2088 | ||
2089 | =item NetBSD | |
2090 | ||
2091 | =item FreeBSD | |
2092 | ||
2d8e9a35 CBW |
2093 | =item Debian GNU/kFreeBSD |
2094 | ||
bb377ba2 JV |
2095 | =item Haiku |
2096 | ||
2097 | =item Irix (6.5. What else?) | |
2098 | ||
2099 | =item OpenBSD | |
2100 | ||
2101 | =item Dragonfly BSD | |
2102 | ||
b60fc215 CBW |
2103 | =item Midnight BSD |
2104 | ||
a62bfce3 CBW |
2105 | =item QNX Neutrino RTOS (6.5.0) |
2106 | ||
bb377ba2 JV |
2107 | =item MirOS BSD |
2108 | ||
7d4dfb6d PG |
2109 | =item Stratus OpenVOS (17.0 or later) |
2110 | ||
bb377ba2 JV |
2111 | Caveats: |
2112 | ||
2113 | =over | |
2114 | ||
2115 | =item time_t issues that may or may not be fixed | |
2116 | ||
2117 | =back | |
2118 | ||
bb377ba2 JV |
2119 | =item Symbian (Series 60 v3, 3.2 and 5 - what else?) |
2120 | ||
10fb90aa | 2121 | =item Stratus VOS / OpenVOS |
bb377ba2 JV |
2122 | |
2123 | =item AIX | |
2124 | ||
25be0a81 BF |
2125 | =item Android |
2126 | ||
dcfa7505 TC |
2127 | =item FreeMINT |
2128 | ||
2129 | Perl now builds with FreeMiNT/Atari. It fails a few tests, that needs | |
2130 | some investigation. | |
2131 | ||
2132 | The FreeMiNT port uses GNU dld for loadable module capabilities. So | |
2133 | ensure you have that library installed when building perl. | |
2134 | ||
bb377ba2 JV |
2135 | =back |
2136 | ||
2890cc8c | 2137 | =head1 EOL Platforms |
7b0e9f13 AD |
2138 | |
2139 | =head2 (Perl 5.20) | |
bb377ba2 JV |
2140 | |
2141 | The following platforms were supported by a previous version of | |
2142 | Perl but have been officially removed from Perl's source code | |
7b0e9f13 | 2143 | as of 5.20: |
bb377ba2 JV |
2144 | |
2145 | =over | |
2146 | ||
7b0e9f13 | 2147 | =item AT&T 3b1 |
bb377ba2 JV |
2148 | |
2149 | =back | |
2150 | ||
7b0e9f13 AD |
2151 | =head2 (Perl 5.14) |
2152 | ||
8cbe99e5 JD |
2153 | The following platforms were supported up to 5.10. They may still |
2154 | have worked in 5.12, but supporting code has been removed for 5.14: | |
bb377ba2 JV |
2155 | |
2156 | =over | |
2157 | ||
2158 | =item Windows 95 | |
2159 | ||
2160 | =item Windows 98 | |
2161 | ||
2162 | =item Windows ME | |
2163 | ||
2164 | =item Windows NT4 | |
2165 | ||
2166 | =back | |
2167 | ||
7b0e9f13 AD |
2168 | =head2 (Perl 5.12) |
2169 | ||
2170 | The following platforms were supported by a previous version of | |
2171 | Perl but have been officially removed from Perl's source code | |
2172 | as of 5.12: | |
2173 | ||
2174 | =over | |
2175 | ||
2176 | =item Atari MiNT | |
2177 | ||
2178 | =item Apollo Domain/OS | |
2179 | ||
2180 | =item Apple Mac OS 8/9 | |
2181 | ||
2182 | =item Tenon Machten | |
2183 | ||
2184 | =back | |
2185 | ||
2186 | ||
bb377ba2 JV |
2187 | =head1 Supported Platforms (Perl 5.8) |
2188 | ||
2189 | As of July 2002 (the Perl release 5.8.0), the following platforms were | |
cec2c193 | 2190 | able to build Perl from the standard source code distribution |
500f1b69 | 2191 | available at L<http://www.cpan.org/src/> |
cec2c193 JH |
2192 | |
2193 | AIX | |
2194 | BeOS | |
6f683aa2 | 2195 | BSD/OS (BSDi) |
cec2c193 | 2196 | Cygwin |
ea297d26 | 2197 | DG/UX |
811b48f2 | 2198 | DOS DJGPP 1) |
cec2c193 JH |
2199 | DYNIX/ptx |
2200 | EPOC R5 | |
2201 | FreeBSD | |
6f683aa2 | 2202 | HI-UXMPP (Hitachi) (5.8.0 worked but we didn't know it) |
cec2c193 JH |
2203 | HP-UX |
2204 | IRIX | |
2205 | Linux | |
8939ba94 | 2206 | Mac OS Classic |
6f683aa2 | 2207 | Mac OS X (Darwin) |
cec2c193 JH |
2208 | MPE/iX |
2209 | NetBSD | |
2210 | NetWare | |
2211 | NonStop-UX | |
6f683aa2 | 2212 | ReliantUNIX (formerly SINIX) |
cec2c193 | 2213 | OpenBSD |
6f683aa2 | 2214 | OpenVMS (formerly VMS) |
3ebac25b | 2215 | Open UNIX (Unixware) (since Perl 5.8.1/5.9.0) |
cec2c193 | 2216 | OS/2 |
522b859a | 2217 | OS/400 (using the PASE) (since Perl 5.8.1/5.9.0) |
70de81db | 2218 | PowerUX |
6f683aa2 | 2219 | POSIX-BC (formerly BS2000) |
cec2c193 JH |
2220 | QNX |
2221 | Solaris | |
70de81db | 2222 | SunOS 4 |
6f683aa2 JH |
2223 | SUPER-UX (NEC) |
2224 | Tru64 UNIX (formerly DEC OSF/1, Digital UNIX) | |
cec2c193 JH |
2225 | UNICOS |
2226 | UNICOS/mk | |
2227 | UTS | |
7d4dfb6d | 2228 | VOS / OpenVOS |
811b48f2 | 2229 | Win95/98/ME/2K/XP 2) |
c40b5d1d | 2230 | WinCE |
6f683aa2 | 2231 | z/OS (formerly OS/390) |
cec2c193 | 2232 | VM/ESA |
ba58ab26 | 2233 | |
811b48f2 JH |
2234 | 1) in DOS mode either the DOS or OS/2 ports can be used |
2235 | 2) compilers: Borland, MinGW (GCC), VC6 | |
cec2c193 | 2236 | |
c40b5d1d | 2237 | The following platforms worked with the previous releases (5.6 and |
cec2c193 JH |
2238 | 5.7), but we did not manage either to fix or to test these in time |
2239 | for the 5.8.0 release. There is a very good chance that many of these | |
70de81db | 2240 | will work fine with the 5.8.0. |
cec2c193 | 2241 | |
8da2b1be | 2242 | BSD/OS |
cec2c193 JH |
2243 | DomainOS |
2244 | Hurd | |
2245 | LynxOS | |
2246 | MachTen | |
2247 | PowerMAX | |
2248 | SCO SV | |
cec2c193 JH |
2249 | SVR4 |
2250 | Unixware | |
2251 | Windows 3.1 | |
ba58ab26 | 2252 | |
70de81db JH |
2253 | Known to be broken for 5.8.0 (but 5.6.1 and 5.7.2 can be used): |
2254 | ||
61988e87 | 2255 | AmigaOS 3 |
70de81db | 2256 | |
ba58ab26 | 2257 | The following platforms have been known to build Perl from source in |
fd46a41b JH |
2258 | the past (5.005_03 and earlier), but we haven't been able to verify |
2259 | their status for the current release, either because the | |
2260 | hardware/software platforms are rare or because we don't have an | |
2261 | active champion on these platforms--or both. They used to work, | |
2262 | though, so go ahead and try compiling them, and let perlbug@perl.org | |
2263 | of any trouble. | |
ba58ab26 | 2264 | |
cec2c193 JH |
2265 | 3b1 |
2266 | A/UX | |
cec2c193 JH |
2267 | ConvexOS |
2268 | CX/UX | |
2269 | DC/OSx | |
2270 | DDE SMES | |
2271 | DOS EMX | |
2272 | Dynix | |
2273 | EP/IX | |
2274 | ESIX | |
2275 | FPS | |
2276 | GENIX | |
2277 | Greenhills | |
2278 | ISC | |
2279 | MachTen 68k | |
cec2c193 JH |
2280 | MPC |
2281 | NEWS-OS | |
2282 | NextSTEP | |
2283 | OpenSTEP | |
2284 | Opus | |
2285 | Plan 9 | |
cec2c193 | 2286 | RISC/os |
8da2b1be | 2287 | SCO ODT/OSR |
cec2c193 JH |
2288 | Stellar |
2289 | SVR2 | |
2290 | TI1500 | |
2291 | TitanOS | |
2292 | Ultrix | |
2293 | Unisys Dynix | |
ba58ab26 JH |
2294 | |
2295 | The following platforms have their own source code distributions and | |
500f1b69 | 2296 | binaries available via L<http://www.cpan.org/ports/> |
ba58ab26 | 2297 | |
cec2c193 | 2298 | Perl release |
ba58ab26 | 2299 | |
522b859a | 2300 | OS/400 (ILE) 5.005_02 |
cec2c193 | 2301 | Tandem Guardian 5.004 |
ba58ab26 JH |
2302 | |
2303 | The following platforms have only binaries available via | |
500f1b69 | 2304 | L<http://www.cpan.org/ports/index.html> : |
ba58ab26 | 2305 | |
cec2c193 | 2306 | Perl release |
ba58ab26 | 2307 | |
cec2c193 JH |
2308 | Acorn RISCOS 5.005_02 |
2309 | AOS 5.002 | |
2310 | LynxOS 5.004_02 | |
ba58ab26 JH |
2311 | |
2312 | Although we do suggest that you always build your own Perl from | |
2313 | the source code, both for maximal configurability and for security, | |
2314 | in case you are in a hurry you can check | |
500f1b69 | 2315 | L<http://www.cpan.org/ports/index.html> for binary distributions. |
ba58ab26 | 2316 | |
c997b287 GS |
2317 | =head1 SEE ALSO |
2318 | ||
b6c36746 | 2319 | L<perlaix>, L<perlamiga>, L<perlbs2000>, |
8d87852b | 2320 | L<perlce>, L<perlcygwin>, L<perldos>, |
469e7be4 | 2321 | L<perlebcdic>, L<perlfreebsd>, L<perlhurd>, L<perlhpux>, L<perlirix>, |
b5afd346 | 2322 | L<perlmacos>, L<perlmacosx>, |
522b859a JH |
2323 | L<perlnetware>, L<perlos2>, L<perlos390>, L<perlos400>, |
2324 | L<perlplan9>, L<perlqnx>, L<perlsolaris>, L<perltru64>, | |
043fec90 | 2325 | L<perlunicode>, L<perlvms>, L<perlvos>, L<perlwin32>, and L<Win32>. |
c997b287 | 2326 | |
e41182b5 GS |
2327 | =head1 AUTHORS / CONTRIBUTORS |
2328 | ||
06e9666b | 2329 | Abigail <abigail@foad.org>, |
c47ff5f1 GS |
2330 | Charles Bailey <bailey@newman.upenn.edu>, |
2331 | Graham Barr <gbarr@pobox.com>, | |
2332 | Tom Christiansen <tchrist@perl.com>, | |
06e9666b | 2333 | Nicholas Clark <nick@ccl4.org>, |
c47ff5f1 | 2334 | Thomas Dorner <Thomas.Dorner@start.de>, |
06e9666b A |
2335 | Andy Dougherty <doughera@lafayette.edu>, |
2336 | Dominic Dunlop <domo@computer.org>, | |
2337 | Neale Ferguson <neale@vma.tabnsw.com.au>, | |
c47ff5f1 | 2338 | David J. Fiander <davidf@mks.com>, |
3fd80bd6 | 2339 | Paul Green <Paul.Green@stratus.com>, |
06e9666b | 2340 | M.J.T. Guy <mjtg@cam.ac.uk>, |
61f30a5e | 2341 | Jarkko Hietaniemi <jhi@iki.fi>, |
c47ff5f1 | 2342 | Luther Huffman <lutherh@stratcom.com>, |
06e9666b A |
2343 | Nick Ing-Simmons <nick@ing-simmons.net>, |
2344 | Andreas J. KE<ouml>nig <a.koenig@mind.de>, | |
c47ff5f1 GS |
2345 | Markus Laker <mlaker@contax.co.uk>, |
2346 | Andrew M. Langmead <aml@world.std.com>, | |
2347 | Larry Moore <ljmoore@freespace.net>, | |
2348 | Paul Moore <Paul.Moore@uk.origin-it.com>, | |
2349 | Chris Nandor <pudge@pobox.com>, | |
1afc07ec | 2350 | Matthias Neeracher <neeracher@mac.com>, |
e71a7dc8 | 2351 | Philip Newton <pne@cpan.org>, |
c47ff5f1 GS |
2352 | Gary Ng <71564.1743@CompuServe.COM>, |
2353 | Tom Phoenix <rootbeer@teleport.com>, | |
2354 | AndrE<eacute> Pirard <A.Pirard@ulg.ac.be>, | |
2355 | Peter Prymmer <pvhp@forte.com>, | |
2356 | Hugo van der Sanden <hv@crypt0.demon.co.uk>, | |
2357 | Gurusamy Sarathy <gsar@activestate.com>, | |
2358 | Paul J. Schinder <schinder@pobox.com>, | |
2359 | Michael G Schwern <schwern@pobox.com>, | |
06e9666b | 2360 | Dan Sugalski <dan@sidhe.org>, |
bbe548ff | 2361 | Nathan Torkington <gnat@frii.com>, |
016930a6 | 2362 | John Malmberg <wb8tyw@qsl.net> |