Commit | Line | Data |
---|---|---|
920dae64 AT |
1 | =head1 NAME |
2 | X<warning, lexical> X<warnings> X<warning> | |
3 | ||
4 | perllexwarn - Perl Lexical Warnings | |
5 | ||
6 | =head1 DESCRIPTION | |
7 | ||
8 | The C<use warnings> pragma is a replacement for both the command line | |
9 | flag B<-w> and the equivalent Perl variable, C<$^W>. | |
10 | ||
11 | The pragma works just like the existing "strict" pragma. | |
12 | This means that the scope of the warning pragma is limited to the | |
13 | enclosing block. It also means that the pragma setting will not | |
14 | leak across files (via C<use>, C<require> or C<do>). This allows | |
15 | authors to independently define the degree of warning checks that will | |
16 | be applied to their module. | |
17 | ||
18 | By default, optional warnings are disabled, so any legacy code that | |
19 | doesn't attempt to control the warnings will work unchanged. | |
20 | ||
21 | All warnings are enabled in a block by either of these: | |
22 | ||
23 | use warnings; | |
24 | use warnings 'all'; | |
25 | ||
26 | Similarly all warnings are disabled in a block by either of these: | |
27 | ||
28 | no warnings; | |
29 | no warnings 'all'; | |
30 | ||
31 | For example, consider the code below: | |
32 | ||
33 | use warnings; | |
34 | my @a; | |
35 | { | |
36 | no warnings; | |
37 | my $b = @a[0]; | |
38 | } | |
39 | my $c = @a[0]; | |
40 | ||
41 | The code in the enclosing block has warnings enabled, but the inner | |
42 | block has them disabled. In this case that means the assignment to the | |
43 | scalar C<$c> will trip the C<"Scalar value @a[0] better written as $a[0]"> | |
44 | warning, but the assignment to the scalar C<$b> will not. | |
45 | ||
46 | =head2 Default Warnings and Optional Warnings | |
47 | ||
48 | Before the introduction of lexical warnings, Perl had two classes of | |
49 | warnings: mandatory and optional. | |
50 | ||
51 | As its name suggests, if your code tripped a mandatory warning, you | |
52 | would get a warning whether you wanted it or not. | |
53 | For example, the code below would always produce an C<"isn't numeric"> | |
54 | warning about the "2:". | |
55 | ||
56 | my $a = "2:" + 3; | |
57 | ||
58 | With the introduction of lexical warnings, mandatory warnings now become | |
59 | I<default> warnings. The difference is that although the previously | |
60 | mandatory warnings are still enabled by default, they can then be | |
61 | subsequently enabled or disabled with the lexical warning pragma. For | |
62 | example, in the code below, an C<"isn't numeric"> warning will only | |
63 | be reported for the C<$a> variable. | |
64 | ||
65 | my $a = "2:" + 3; | |
66 | no warnings; | |
67 | my $b = "2:" + 3; | |
68 | ||
69 | Note that neither the B<-w> flag or the C<$^W> can be used to | |
70 | disable/enable default warnings. They are still mandatory in this case. | |
71 | ||
72 | =head2 What's wrong with B<-w> and C<$^W> | |
73 | ||
74 | Although very useful, the big problem with using B<-w> on the command | |
75 | line to enable warnings is that it is all or nothing. Take the typical | |
76 | scenario when you are writing a Perl program. Parts of the code you | |
77 | will write yourself, but it's very likely that you will make use of | |
78 | pre-written Perl modules. If you use the B<-w> flag in this case, you | |
79 | end up enabling warnings in pieces of code that you haven't written. | |
80 | ||
81 | Similarly, using C<$^W> to either disable or enable blocks of code is | |
82 | fundamentally flawed. For a start, say you want to disable warnings in | |
83 | a block of code. You might expect this to be enough to do the trick: | |
84 | ||
85 | { | |
86 | local ($^W) = 0; | |
87 | my $a =+ 2; | |
88 | my $b; chop $b; | |
89 | } | |
90 | ||
91 | When this code is run with the B<-w> flag, a warning will be produced | |
92 | for the C<$a> line -- C<"Reversed += operator">. | |
93 | ||
94 | The problem is that Perl has both compile-time and run-time warnings. To | |
95 | disable compile-time warnings you need to rewrite the code like this: | |
96 | ||
97 | { | |
98 | BEGIN { $^W = 0 } | |
99 | my $a =+ 2; | |
100 | my $b; chop $b; | |
101 | } | |
102 | ||
103 | The other big problem with C<$^W> is the way you can inadvertently | |
104 | change the warning setting in unexpected places in your code. For example, | |
105 | when the code below is run (without the B<-w> flag), the second call | |
106 | to C<doit> will trip a C<"Use of uninitialized value"> warning, whereas | |
107 | the first will not. | |
108 | ||
109 | sub doit | |
110 | { | |
111 | my $b; chop $b; | |
112 | } | |
113 | ||
114 | doit(); | |
115 | ||
116 | { | |
117 | local ($^W) = 1; | |
118 | doit() | |
119 | } | |
120 | ||
121 | This is a side-effect of C<$^W> being dynamically scoped. | |
122 | ||
123 | Lexical warnings get around these limitations by allowing finer control | |
124 | over where warnings can or can't be tripped. | |
125 | ||
126 | =head2 Controlling Warnings from the Command Line | |
127 | ||
128 | There are three Command Line flags that can be used to control when | |
129 | warnings are (or aren't) produced: | |
130 | ||
131 | =over 5 | |
132 | ||
133 | =item B<-w> | |
134 | X<-w> | |
135 | ||
136 | This is the existing flag. If the lexical warnings pragma is B<not> | |
137 | used in any of you code, or any of the modules that you use, this flag | |
138 | will enable warnings everywhere. See L<Backward Compatibility> for | |
139 | details of how this flag interacts with lexical warnings. | |
140 | ||
141 | =item B<-W> | |
142 | X<-W> | |
143 | ||
144 | If the B<-W> flag is used on the command line, it will enable all warnings | |
145 | throughout the program regardless of whether warnings were disabled | |
146 | locally using C<no warnings> or C<$^W =0>. This includes all files that get | |
147 | included via C<use>, C<require> or C<do>. | |
148 | Think of it as the Perl equivalent of the "lint" command. | |
149 | ||
150 | =item B<-X> | |
151 | X<-X> | |
152 | ||
153 | Does the exact opposite to the B<-W> flag, i.e. it disables all warnings. | |
154 | ||
155 | =back | |
156 | ||
157 | =head2 Backward Compatibility | |
158 | ||
159 | If you are used with working with a version of Perl prior to the | |
160 | introduction of lexically scoped warnings, or have code that uses both | |
161 | lexical warnings and C<$^W>, this section will describe how they interact. | |
162 | ||
163 | How Lexical Warnings interact with B<-w>/C<$^W>: | |
164 | ||
165 | =over 5 | |
166 | ||
167 | =item 1. | |
168 | ||
169 | If none of the three command line flags (B<-w>, B<-W> or B<-X>) that | |
170 | control warnings is used and neither C<$^W> or the C<warnings> pragma | |
171 | are used, then default warnings will be enabled and optional warnings | |
172 | disabled. | |
173 | This means that legacy code that doesn't attempt to control the warnings | |
174 | will work unchanged. | |
175 | ||
176 | =item 2. | |
177 | ||
178 | The B<-w> flag just sets the global C<$^W> variable as in 5.005 -- this | |
179 | means that any legacy code that currently relies on manipulating C<$^W> | |
180 | to control warning behavior will still work as is. | |
181 | ||
182 | =item 3. | |
183 | ||
184 | Apart from now being a boolean, the C<$^W> variable operates in exactly | |
185 | the same horrible uncontrolled global way, except that it cannot | |
186 | disable/enable default warnings. | |
187 | ||
188 | =item 4. | |
189 | ||
190 | If a piece of code is under the control of the C<warnings> pragma, | |
191 | both the C<$^W> variable and the B<-w> flag will be ignored for the | |
192 | scope of the lexical warning. | |
193 | ||
194 | =item 5. | |
195 | ||
196 | The only way to override a lexical warnings setting is with the B<-W> | |
197 | or B<-X> command line flags. | |
198 | ||
199 | =back | |
200 | ||
201 | The combined effect of 3 & 4 is that it will allow code which uses | |
202 | the C<warnings> pragma to control the warning behavior of $^W-type | |
203 | code (using a C<local $^W=0>) if it really wants to, but not vice-versa. | |
204 | ||
205 | =head2 Category Hierarchy | |
206 | X<warning, categories> | |
207 | ||
208 | A hierarchy of "categories" have been defined to allow groups of warnings | |
209 | to be enabled/disabled in isolation. | |
210 | ||
211 | The current hierarchy is: | |
212 | ||
213 | all -+ | |
214 | | | |
215 | +- closure | |
216 | | | |
217 | +- deprecated | |
218 | | | |
219 | +- exiting | |
220 | | | |
221 | +- glob | |
222 | | | |
223 | +- io -----------+ | |
224 | | | | |
225 | | +- closed | |
226 | | | | |
227 | | +- exec | |
228 | | | | |
229 | | +- layer | |
230 | | | | |
231 | | +- newline | |
232 | | | | |
233 | | +- pipe | |
234 | | | | |
235 | | +- unopened | |
236 | | | |
237 | +- misc | |
238 | | | |
239 | +- numeric | |
240 | | | |
241 | +- once | |
242 | | | |
243 | +- overflow | |
244 | | | |
245 | +- pack | |
246 | | | |
247 | +- portable | |
248 | | | |
249 | +- recursion | |
250 | | | |
251 | +- redefine | |
252 | | | |
253 | +- regexp | |
254 | | | |
255 | +- severe -------+ | |
256 | | | | |
257 | | +- debugging | |
258 | | | | |
259 | | +- inplace | |
260 | | | | |
261 | | +- internal | |
262 | | | | |
263 | | +- malloc | |
264 | | | |
265 | +- signal | |
266 | | | |
267 | +- substr | |
268 | | | |
269 | +- syntax -------+ | |
270 | | | | |
271 | | +- ambiguous | |
272 | | | | |
273 | | +- bareword | |
274 | | | | |
275 | | +- digit | |
276 | | | | |
277 | | +- parenthesis | |
278 | | | | |
279 | | +- precedence | |
280 | | | | |
281 | | +- printf | |
282 | | | | |
283 | | +- prototype | |
284 | | | | |
285 | | +- qw | |
286 | | | | |
287 | | +- reserved | |
288 | | | | |
289 | | +- semicolon | |
290 | | | |
291 | +- taint | |
292 | | | |
293 | +- threads | |
294 | | | |
295 | +- uninitialized | |
296 | | | |
297 | +- unpack | |
298 | | | |
299 | +- untie | |
300 | | | |
301 | +- utf8 | |
302 | | | |
303 | +- void | |
304 | | | |
305 | +- y2k | |
306 | ||
307 | Just like the "strict" pragma any of these categories can be combined | |
308 | ||
309 | use warnings qw(void redefine); | |
310 | no warnings qw(io syntax untie); | |
311 | ||
312 | Also like the "strict" pragma, if there is more than one instance of the | |
313 | C<warnings> pragma in a given scope the cumulative effect is additive. | |
314 | ||
315 | use warnings qw(void); # only "void" warnings enabled | |
316 | ... | |
317 | use warnings qw(io); # only "void" & "io" warnings enabled | |
318 | ... | |
319 | no warnings qw(void); # only "io" warnings enabled | |
320 | ||
321 | To determine which category a specific warning has been assigned to see | |
322 | L<perldiag>. | |
323 | ||
324 | Note: In Perl 5.6.1, the lexical warnings category "deprecated" was a | |
325 | sub-category of the "syntax" category. It is now a top-level category | |
326 | in its own right. | |
327 | ||
328 | ||
329 | =head2 Fatal Warnings | |
330 | X<warning, fatal> | |
331 | ||
332 | The presence of the word "FATAL" in the category list will escalate any | |
333 | warnings detected from the categories specified in the lexical scope | |
334 | into fatal errors. In the code below, the use of C<time>, C<length> | |
335 | and C<join> can all produce a C<"Useless use of xxx in void context"> | |
336 | warning. | |
337 | ||
338 | use warnings; | |
339 | ||
340 | time; | |
341 | ||
342 | { | |
343 | use warnings FATAL => qw(void); | |
344 | length "abc"; | |
345 | } | |
346 | ||
347 | join "", 1,2,3; | |
348 | ||
349 | print "done\n"; | |
350 | ||
351 | When run it produces this output | |
352 | ||
353 | Useless use of time in void context at fatal line 3. | |
354 | Useless use of length in void context at fatal line 7. | |
355 | ||
356 | The scope where C<length> is used has escalated the C<void> warnings | |
357 | category into a fatal error, so the program terminates immediately it | |
358 | encounters the warning. | |
359 | ||
360 | To explicitly turn off a "FATAL" warning you just disable the warning | |
361 | it is associated with. So, for example, to disable the "void" warning | |
362 | in the example above, either of these will do the trick: | |
363 | ||
364 | no warnings qw(void); | |
365 | no warnings FATAL => qw(void); | |
366 | ||
367 | If you want to downgrade a warning that has been escalated into a fatal | |
368 | error back to a normal warning, you can use the "NONFATAL" keyword. For | |
369 | example, the code below will promote all warnings into fatal errors, | |
370 | except for those in the "syntax" category. | |
371 | ||
372 | use warnings FATAL => 'all', NONFATAL => 'syntax'; | |
373 | ||
374 | =head2 Reporting Warnings from a Module | |
375 | X<warning, reporting> X<warning, registering> | |
376 | ||
377 | The C<warnings> pragma provides a number of functions that are useful for | |
378 | module authors. These are used when you want to report a module-specific | |
379 | warning to a calling module has enabled warnings via the C<warnings> | |
380 | pragma. | |
381 | ||
382 | Consider the module C<MyMod::Abc> below. | |
383 | ||
384 | package MyMod::Abc; | |
385 | ||
386 | use warnings::register; | |
387 | ||
388 | sub open { | |
389 | my $path = shift; | |
390 | if ($path !~ m#^/#) { | |
391 | warnings::warn("changing relative path to /var/abc") | |
392 | if warnings::enabled(); | |
393 | $path = "/var/abc/$path"; | |
394 | } | |
395 | } | |
396 | ||
397 | 1; | |
398 | ||
399 | The call to C<warnings::register> will create a new warnings category | |
400 | called "MyMod::abc", i.e. the new category name matches the current | |
401 | package name. The C<open> function in the module will display a warning | |
402 | message if it gets given a relative path as a parameter. This warnings | |
403 | will only be displayed if the code that uses C<MyMod::Abc> has actually | |
404 | enabled them with the C<warnings> pragma like below. | |
405 | ||
406 | use MyMod::Abc; | |
407 | use warnings 'MyMod::Abc'; | |
408 | ... | |
409 | abc::open("../fred.txt"); | |
410 | ||
411 | It is also possible to test whether the pre-defined warnings categories are | |
412 | set in the calling module with the C<warnings::enabled> function. Consider | |
413 | this snippet of code: | |
414 | ||
415 | package MyMod::Abc; | |
416 | ||
417 | sub open { | |
418 | warnings::warnif("deprecated", | |
419 | "open is deprecated, use new instead"); | |
420 | new(@_); | |
421 | } | |
422 | ||
423 | sub new | |
424 | ... | |
425 | 1; | |
426 | ||
427 | The function C<open> has been deprecated, so code has been included to | |
428 | display a warning message whenever the calling module has (at least) the | |
429 | "deprecated" warnings category enabled. Something like this, say. | |
430 | ||
431 | use warnings 'deprecated'; | |
432 | use MyMod::Abc; | |
433 | ... | |
434 | MyMod::Abc::open($filename); | |
435 | ||
436 | Either the C<warnings::warn> or C<warnings::warnif> function should be | |
437 | used to actually display the warnings message. This is because they can | |
438 | make use of the feature that allows warnings to be escalated into fatal | |
439 | errors. So in this case | |
440 | ||
441 | use MyMod::Abc; | |
442 | use warnings FATAL => 'MyMod::Abc'; | |
443 | ... | |
444 | MyMod::Abc::open('../fred.txt'); | |
445 | ||
446 | the C<warnings::warnif> function will detect this and die after | |
447 | displaying the warning message. | |
448 | ||
449 | The three warnings functions, C<warnings::warn>, C<warnings::warnif> | |
450 | and C<warnings::enabled> can optionally take an object reference in place | |
451 | of a category name. In this case the functions will use the class name | |
452 | of the object as the warnings category. | |
453 | ||
454 | Consider this example: | |
455 | ||
456 | package Original; | |
457 | ||
458 | no warnings; | |
459 | use warnings::register; | |
460 | ||
461 | sub new | |
462 | { | |
463 | my $class = shift; | |
464 | bless [], $class; | |
465 | } | |
466 | ||
467 | sub check | |
468 | { | |
469 | my $self = shift; | |
470 | my $value = shift; | |
471 | ||
472 | if ($value % 2 && warnings::enabled($self)) | |
473 | { warnings::warn($self, "Odd numbers are unsafe") } | |
474 | } | |
475 | ||
476 | sub doit | |
477 | { | |
478 | my $self = shift; | |
479 | my $value = shift; | |
480 | $self->check($value); | |
481 | # ... | |
482 | } | |
483 | ||
484 | 1; | |
485 | ||
486 | package Derived; | |
487 | ||
488 | use warnings::register; | |
489 | use Original; | |
490 | our @ISA = qw( Original ); | |
491 | sub new | |
492 | { | |
493 | my $class = shift; | |
494 | bless [], $class; | |
495 | } | |
496 | ||
497 | ||
498 | 1; | |
499 | ||
500 | The code below makes use of both modules, but it only enables warnings from | |
501 | C<Derived>. | |
502 | ||
503 | use Original; | |
504 | use Derived; | |
505 | use warnings 'Derived'; | |
506 | my $a = new Original; | |
507 | $a->doit(1); | |
508 | my $b = new Derived; | |
509 | $a->doit(1); | |
510 | ||
511 | When this code is run only the C<Derived> object, C<$b>, will generate | |
512 | a warning. | |
513 | ||
514 | Odd numbers are unsafe at main.pl line 7 | |
515 | ||
516 | Notice also that the warning is reported at the line where the object is first | |
517 | used. | |
518 | ||
519 | =head1 TODO | |
520 | ||
521 | perl5db.pl | |
522 | The debugger saves and restores C<$^W> at runtime. I haven't checked | |
523 | whether the debugger will still work with the lexical warnings | |
524 | patch applied. | |
525 | ||
526 | diagnostics.pm | |
527 | I *think* I've got diagnostics to work with the lexical warnings | |
528 | patch, but there were design decisions made in diagnostics to work | |
529 | around the limitations of C<$^W>. Now that those limitations are gone, | |
530 | the module should be revisited. | |
531 | ||
532 | document calling the warnings::* functions from XS | |
533 | ||
534 | =head1 SEE ALSO | |
535 | ||
536 | L<warnings>, L<perldiag>. | |
537 | ||
538 | =head1 AUTHOR | |
539 | ||
540 | Paul Marquess |