1HTML::Mason::Admin(3) User Contributed Perl DocumentationHTML::Mason::Admin(3)
2
3
4
6 HTML::Mason::Admin - Mason Administrator's Manual
7
9 This manual is written for the sysadmin/webmaster in charge of
10 installing, configuring, or tuning a Mason system. The bulk of the
11 documentation assumes that you are using mod_perl. See RUNNING OUTSIDE
12 OF MOD_PERL for more details. For more details on mod_perl, visit the
13 mod_perl website at http://perl.apache.org/.
14
16 Mason includes a module specifically designed to integrate Mason and
17 mod_perl (1 and 2), "HTML::Mason::ApacheHandler". By telling mod_perl
18 to hand content requests to this module, you can use Mason to generate
19 web pages. There are two ways to configure Mason under mod_perl.
20
21 · Basic
22
23 Mason provides reasonable default behavior under mod_perl, so using
24 Mason can be as simple as adding two directives to your Apache
25 configuration file. Throughout this document, we will assume that
26 your Apache configuration file is called httpd.conf. By adding
27 more configuration parameters to this file you can implement more
28 complex behaviors.
29
30 · Advanced
31
32 If the basic method does not provide enough flexibility for you,
33 you can wrap Mason in a custom mod_perl handler. The wrapper code
34 you write can create its own Mason objects, or it can take
35 advantage of httpd.conf configuration parameters and let Mason
36 create the objects it needs by itself.
37
38 We recommend that you start with the basic method and work your way
39 forward as the need for flexibility arises.
40
41 Mason is very flexible, and you can replace parts of it by creating
42 your own classes. This documentation assumes that you are simply using
43 the classes provided in the Mason distribution. Subclassing is covered
44 in the Subclassing document. The two topics are orthogonal, as you can
45 mix the configuration techniques discussed here with your own custom
46 subclasses.
47
49 The absolutely most minimal configuration looks like this:
50
51 PerlModule HTML::Mason::ApacheHandler
52
53 <Location />
54 SetHandler perl-script
55 PerlHandler HTML::Mason::ApacheHandler
56 </Location>
57
58 This configuration tells Apache to serve all URLs through Mason (see
59 the next section for a more realistic strategy). We use the PerlModule
60 line to tell mod_perl to load Mason once at startup time, saving time
61 and memory. This example does not set any Mason configuration
62 parameters, so Mason uses its default values.
63
64 If this is your first time installing and using Mason, we recommend
65 that you use the above configuration in a test webserver to start with.
66 This will let you play with Mason under mod_perl with a minimum of
67 fuss. Once you've gotten this working, then come back and read the
68 rest of the document for further possibilities.
69
70 Controlling Access via Filename Extension
71 As it turns out, serving every URL through Mason is a bad idea for two
72 reasons:
73
74 1. Mason should be prevented from handling images, tarballs, and other
75 binary files. Not only will performance suffer, but binary files
76 may inadvertently contain a Mason character sequence such as "<%".
77 These files should be instead served by Apache's default content
78 handler.
79
80 2. Mason should be prevented from serving private (non-top-level)
81 Mason components to users. For example, if you used a utility
82 component for performing arbitrary sql queries, you wouldn't want
83 external users to be able to access it via a URL. Requests for
84 private components should simply result in a 404 NOT_FOUND.
85
86 The easiest way to distinguish between different types of files is with
87 filename extensions. While many naming schemes are possible, we suggest
88 using "normal" extensions for top-level components and adding an "m"
89 prefix for private components. For example,
90
91 Top-level Private
92
93 Component outputs HTML .html .mhtml
94 Component outputs text .txt .mtxt
95 Component executes Perl .pl .mpl
96
97 This scheme minimizes the chance of confusing browsers about content
98 type, scales well for new classes of content (e.g. .js/.mjs for
99 javascript), and makes transparent the fact that you are using Mason
100 versus some other package.
101
102 Here is a configuration that enforces this naming scheme:
103
104 PerlModule HTML::Mason::ApacheHandler
105
106 <LocationMatch "(\.html|\.txt|\.pl)$">
107 SetHandler perl-script
108 PerlHandler HTML::Mason::ApacheHandler
109 </LocationMatch>
110
111 <LocationMatch "(\.m(html|txt|pl)|dhandler|autohandler)$">
112 SetHandler perl-script
113 PerlInitHandler Apache::Constants::NOT_FOUND
114 </LocationMatch>
115
116 The first block causes URLs ending in .html, .txt, or .pl to be served
117 through Mason. The second block causes requests to private components
118 to return 404 NOT_FOUND, preventing unscrupulous users from even
119 knowing which private components exist. Any other file extensions (e.g.
120 .gif, .tgz) will be served by Apache's default content handler.
121
122 You might prefer "FilesMatch" to "LocationMatch". However, be aware
123 that "LocationMatch" will work best in conjunction with Mason's
124 dhandlers.
125
126 Configuration Parameters
127 Mason allows you to flexibly configure its behavior via httpd.conf
128 configuration parameters.
129
130 These configuration parameters are set via mod_perl's "PerlSetVar" and
131 "PerlAddVar" directives. Though these parameters are all strings in
132 your httpd.conf file, Mason treats different directives as containing
133 different types of values:
134
135 · string
136
137 The variable's value is simply taken literally and used. The
138 string should be surrounded by quotes if the it contains
139 whitespace. The quotes will be automatically removed by Apache
140 before Mason sees the variable.
141
142 · boolean
143
144 The variable's value is used as a boolean, and is subject to Perl's
145 rules on truth/falseness. It is recommended that you use 0 (false)
146 or 1 (true) for these arguments.
147
148 · code
149
150 The string is treated as a piece of code and "eval"'ed. This is
151 used for parameters that expect subroutine references. For
152 example, an anonymous subroutine might look like:
153
154 PerlSetVar MasonOutMode "sub { ... }"
155
156 A named subroutine reference would look like this:
157
158 PerlSetVar MasonOutMode "\&Some::Module::handle_output"
159
160 · list
161
162 To set a list parameter, use "PerlAddVar" for the values, like
163 this:
164
165 PerlAddVar MasonPreloads /foo/bar/baz.comp
166 PerlAddVar MasonPreloads /foo/bar/quux.comp
167
168 · hash_list
169
170 Just like a list parameter, use "PerlAddVar" for the values.
171 However, in the case of a hash_list, each element should be a
172 key/value pair separated by "=>":
173
174 PerlAddVar MasonDataCacheDefaults "cache_class => MemoryCache"
175 PerlAddVar MasonDataCacheDefaults "namespace => foo"
176
177 Take note that the right hand side of the each pair should not be
178 quoted.
179
180 See HTML::Mason::Params for a full list of parameters, and their
181 associated types.
182
184 Component Root
185 The component root (comp_root) marks the top of your component
186 hierarchy. When running Mason with the ApacheHandler or CGIHandler
187 modules, this defaults to your document root.
188
189 The component root defines how component paths are translated into real
190 file paths. If your component root is /usr/local/httpd/docs, a
191 component path of /products/index.html translates to the file
192 /usr/local/httpd/docs/products/index.html.
193
194 One cannot call a component outside the component root. If Apache
195 passes a file through Mason that is outside the component root (say, as
196 the result of an Alias) you will get a 404 and a warning in the logs.
197
198 You may also specify multiple component roots in the spirit of Perl's
199 @INC. Each root is assigned a key that identifies the root
200 mnemonically. For example, in httpd.conf:
201
202 PerlAddVar MasonCompRoot "private => /usr/home/joe/comps"
203 PerlAddVar MasonCompRoot "main => /usr/local/www/htdocs"
204
205 This specifies two component roots, a main component tree and a private
206 tree which overrides certain components. The order is respected ala
207 @INC, so private is searched first and main second.
208
209 The component root keys must be unique in a case-insensitive
210 comparison. The keys are used in several ways. They help to distinguish
211 component caches and object files between different component roots,
212 and they appear in the "title()" of a component.
213
214 Data Directory
215 The data directory (data_dir) is a writable directory that Mason uses
216 for various features and optimizations. By default, it is a directory
217 called "mason" under your Apache server root. Because Mason will not
218 use a default data directory under a top-level directory, you will need
219 to change this on certain systems that assign a high-level server root
220 such as /usr or /etc.
221
222 Mason will create the directory on startup, if necessary, and set its
223 permissions according to the web server User/Group.
224
225 External Modules
226 Components will often need access to external Perl modules. There are
227 several ways to load them.
228
229 · The httpd PerlModule directive:
230
231 PerlModule CGI
232 PerlModule LWP
233
234 · In the "<%once>" section of the component(s) that use the module.
235
236 <%once>
237 use CGI ':standard';
238 use LWP;
239 </%once>
240
241 Each method has its own trade-offs:
242
243 The first method ensures that the module will be loaded by the Apache
244 parent process at startup time, saving time and memory. The second
245 method, in contrast, will cause the modules to be loaded by each server
246 child. On the other hand this could save memory if the component and
247 module are rarely used. See the mod_perl guide's tuning section and
248 Vivek Khera's mod_perl tuning guide for more details on this issue.
249
250 The second method uses the modules from inside the package used by
251 components ("HTML::Mason::Commands"), meaning that exported method
252 names and other symbols will be usable from components. The first
253 method, in contrast, will import symbols into the "main" package. The
254 significance of this depends on whether the modules export symbols and
255 whether you want to use them from components.
256
257 If you want to preload the modules in your httpd.conf file, and still
258 have them export symbols into the "HTML::Mason::Commands" namespace,
259 you can do this:
260
261 <Perl>
262 { package HTML::Mason::Commands;
263 use CGI;
264 use LWP;
265 }
266 </Perl>
267
268 A Perl section will also work for including local library paths:
269
270 <Perl>
271 use lib '/path/to/local/lib';
272 </Perl>
273
274 Allowing Directory Requests
275 By default Mason will decline requests for directories, leaving Apache
276 to serve up a directory index or a FORBIDDEN as appropriate.
277 Unfortunately this rule applies even if there is a dhandler in the
278 directory: /foo/bar/dhandler does not get a chance to handle a request
279 for /foo/bar/.
280
281 If you would like Mason to handle directory requests, set decline_dirs
282 to 0. The dhandler that catches a directory request is responsible for
283 setting a reasonable content type via "$r->content_type()".
284
285 Configuring Virtual Sites
286 These examples extend the single site configurations given so far.
287
288 Multiple sites, one component root
289
290 If you want to share some components between your sites, arrange your
291 httpd.conf so that all DocumentRoots live under a single component
292 space:
293
294 # Web site #1
295 <VirtualHost www.site1.com>
296 DocumentRoot /usr/local/www/htdocs/site1
297 <LocationMatch ...>
298 SetHandler perl-script
299 PerlHandler HTML::Mason::ApacheHandler
300 </LocationMatch>
301 </VirtualHost>
302
303 # Web site #2
304 <VirtualHost www.site2.com>
305 DocumentRoot /usr/local/www/htdocs/site2
306 <LocationMatch ...>
307 SetHandler perl-script
308 PerlHandler HTML::Mason::ApacheHandler
309 </LocationMatch>
310 </VirtualHost>
311
312 # Mason configuration
313 PerlSetVar MasonCompRoot /usr/local/www/htdocs
314 PerlSetVar MasonDataDir /usr/local/mason
315 PerlModule HTML::Mason::ApacheHandler
316
317 The directory structure for this scenario might look like:
318
319 /usr/local/www/htdocs/ # component root
320 +- shared/ # shared components
321 +- site1/ # DocumentRoot for first site
322 +- site2/ # DocumentRoot for second site
323
324 Incoming URLs for each site can only request components in their
325 respective DocumentRoots, while components internally can call other
326 components anywhere in the component space. The shared/ directory is a
327 private directory for use by components, inaccessible from the Web.
328
329 Multiple sites, multiple component roots
330
331 If your sites need to have completely distinct component hierarchies,
332 e.g. if you are providing Mason ISP services for multiple users, then
333 the component root must change depending on the site requested.
334
335 <VirtualHost www.site1.com>
336 DocumentRoot /usr/local/www/htdocs/site1
337
338 # Mason configuration
339 PerlSetVar MasonCompRoot /usr/local/www/htdocs/site1
340 PerlSetVar MasonDataDir /usr/local/mason/site1
341
342 <LocationMatch ...>
343 SetHandler perl-script
344 PerlHandler HTML::Mason::ApacheHandler
345 </LocationMatch>
346 </VirtualHost>
347
348 # Web site #2
349 <VirtualHost www.site2.com>
350 DocumentRoot /usr/local/www/htdocs/site2
351
352 # Mason configuration
353 PerlSetVar MasonCompRoot /usr/local/www/htdocs/site2
354 PerlSetVar MasonDataDir /usr/local/mason/site2
355
356 <LocationMatch ...>
357 SetHandler perl-script
358 PerlHandler HTML::Mason::ApacheHandler
359 </LocationMatch>
360 </VirtualHost>
361
363 As mentioned previously, it is possible to write a custom mod_perl
364 content handler that wraps around Mason and provides basically
365 unlimited flexibility when handling requests. In this section, we show
366 some basic wrappers and re-implement some of the functionality
367 previously discussed, such as declining image requests and protecting
368 private components.
369
370 In addition, we discuss some of the possibilities that become available
371 when you create a custom wrapper around Mason's request handling
372 mechanism. This wrapper generally consists of two parts. The
373 initialization portion, run at server startup, will load any needed
374 modules and create objects. The other portion is the "handler()"
375 subroutine, which handles web page requests.
376
377 Writing a Wrapper
378 To create a wrapper, you simply need to define a "handler()" subroutine
379 in the package of your choice, and tell mod_perl to use it as a content
380 handler. The file that defines the "handler()" subroutine can be a
381 module, or you can simply load a simple file that contains this
382 subroutine definition. The latter solution was, for a long time, the
383 only way to configure Mason, and the file used was traditionally called
384 handler.pl.
385
386 Nowadays, we recommend that you create a custom module in the
387 appropriate namespace and define your "handler()" subroutine there.
388 The advantage to this approach is that it uses well-known techniques
389 for creating and installing modules, but it does require a bit more
390 work than simply dropping a script file into the Apache configuration
391 directory. But because the process is better defined, it may "feel"
392 more solid to some folks than the script approach.
393
394 The eg/ directory of the Mason distribution contains a couple sample
395 modules that define "handler()" subroutines. Let's assume that your
396 module, like the example, defines a "handler()" in the package
397 "MyApp::Mason". In this case, your Apache configuration would look
398 like this:
399
400 PerlModule MyApp::Mason
401
402 <LocationMatch ...>
403 SetHandler perl-script
404 PerlHandler MyApp::Mason
405 </LocationMatch>
406
407 You may still see references to a handler.pl file in the Mason users
408 list archives, as well as the FAQ. These references will generally be
409 applicable to any custom code wrapping Mason.
410
411 Wrappers and PerlSetVar-style configuration
412
413 Sometimes people attempt to write a wrapper and configure Mason with
414 "PerlSetVar" directives in their Apache configuration file. This does
415 not work. When you give mod_perl this configuration:
416
417 PerlHandler HTML::Mason::ApacheHandler
418
419 it will dispatch directly to the
420 "HTML::Mason::ApacheHandler->handler()" method, without ever executing
421 your wrapper code. However, you can mix the two methods. See Mixing
422 httpd.conf Configuration with a Wrapper
423
424 Wrapping with a <Perl> block
425 You can also put your wrapper code in a "<Perl>" block as part of your
426 httpd.conf file. The result is no different than loading a file via
427 the "PerlRequire" directive.
428
429 The Wrapper Code
430 Regardless of how you load your wrapper code, it will always work the
431 same way. The "handler()" subroutine should expect to receive the
432 Apache request object representing the current request. This request
433 object is used by the ApacheHandler module to determine what component
434 is being called.
435
436 Let's look at the guts of some wrapper code. Here's a first version:
437
438 package MyApp::Mason;
439
440 use strict;
441 use HTML::Mason::ApacheHandler;
442
443 my $ah =
444 HTML::Mason::ApacheHandler->new
445 ( comp_root => '/path/to/comp/root',
446 data_dir => '/path/to/data/dir' );
447
448 sub handler {
449 my ($r) = @_;
450
451 return $ah->handle_request($r);
452 }
453
454 This wrapper is fully functional, but it doesn't actually do anything
455 you couldn't do more easily by configuring Mason via the httpd.conf
456 file. However, it does serve as a good skeleton to which additional
457 functionality can easily be added.
458
459 External Modules Revisited
460 Since you are loading an arbitrary piece of code to define your
461 wrapper, you can easily load other modules needed for your application
462 at the same time. For example, you might simple add these lines to the
463 wrapper code above:
464
465 {
466 package HTML::Mason::Commands;
467
468 use MIME::Base64;
469 }
470
471 Explicitly setting the package to "HTML::Mason::Commands" makes sure
472 that any symbols that the loaded modules export (constants,
473 subroutines, etc.) get exported into the namespace under which
474 components run. Of course, if you've changed the component namespace,
475 make sure to change the package name here as well.
476
477 Alternatively, you might consider creating a separate piece of code to
478 load the modules you need. For example, you might create a module
479 called "MyApp::MasonInit":
480
481 {
482 package HTML::Mason::Commands;
483
484 use Apache::Constants qw(:common);
485 use Apache::URI;
486 use File::Temp;
487 }
488
489 1;
490
491 This can be loaded via a "PerlModule" directive in the httpd.conf file,
492 or in the wrapper code itself via "use".
493
494 Example: Controlling access with component attributes
495
496 An example of something you can only do with wrapper code is deciding
497 at run-time whether a component can be accessed at the top-level based
498 on a complex property of the component. For example, here's a piece of
499 code that uses the current user and a component's "access_level"
500 attribute to control access:
501
502 sub handler {
503 my ($r) = @_;
504
505 my $req = $ah->prepare_request($r);
506
507 my $comp = $req->request_comp;
508
509 # this is done via magic hand-waving ...
510 my $user = get_user_from_cookie();
511
512 # remember, attributes are inherited so this could come from a
513 # component higher up the inheritance chain
514 my $required_access = $comp->attr('access_level');
515
516 return NOT_FOUND
517 if $user->access_level < $required_access;
518
519 return $req->exec;
520 }
521
522 Wrappers with Virtual Hosts
523 If you had several virtual hosts, each of which had a separate
524 component root, you'd need to create a separate ApacheHandler object
525 for each host, one for each host. Here's some sample code for that:
526
527 my %ah;
528 foreach my $site ( qw( site1 site2 site3 ) ) {
529 $ah{$site} =
530 HTML::Mason::ApacheHandler->new
531 ( comp_root => "/usr/local/www/$site",
532 data_dir => "/usr/local/mason/$site" );
533 }
534
535 sub handler {
536 my ($r) = @_;
537
538 my $site = $r->dir_config('SiteName');
539
540 return DECLINED unless exists $ah{$site};
541
542 return $ah{$site}->handle_request($r);
543 }
544
545 This code assumes that you set the "SiteName" variable via a
546 "PerlSetVar" directive in each "VirtualHost" block, like this:
547
548 <VirtualHost site1.example.com>
549 PerlSetVar SiteName site1
550
551 <LocationMatch ...>
552 SetHandler perl-script
553 PerlHandler MyApp::Mason
554 </LocationMatch>
555 </VirtualHost>
556
557 Creating apachehandler objects on the fly
558
559 You might also consider creating ApacheHandler objects on the fly, like
560 this:
561
562 my %ah;
563 sub handler {
564 my ($r) = @_;
565 my $site = $r->dir_config('SiteName');
566
567 return DECLINED unless $site;
568
569 unless exists($ah{$site}) {
570 $ah{$site} = HTML::Mason::ApacheHandler->new( ... );
571 }
572
573 $ah{$site}->handle_request($r);
574 }
575
576 This is more flexible but you lose the memory savings of creating all
577 your objects during server startup.
578
579 Other uses for a wrapper
580
581 If you have some code which must always run after a request, then the
582 only way to guarantee that this happens is to wrap the
583 "$ah->handle_request()" call in an "eval {}" block, and then run the
584 needed code after the request returns. You can then handle errors
585 however you like.
586
587 Mixing httpd.conf Configuration with a Wrapper
588 You can take advantage of Mason's httpd.conf configuration system while
589 at the same time providing your own wrapper code. The key to doing
590 this is not creating your own ApacheHandler object. Instead, you call
591 the "HTML::Mason::ApacheHandler->handler()" class method from your
592 "handler()" subroutine. Here's a complete wrapper that does this:
593
594 package MyApp::Mason;
595
596 use strict;
597 use HTML::Mason::ApacheHandler;
598
599 sub handler {
600 my ($r) = @_;
601
602 return HTML::Mason::ApacheHandler->handler($r);
603 }
604
605 The "HTML::Mason::ApacheHandler->handler" method will create an
606 ApacheHandler object based on the configuration directives it finds in
607 your httpd.conf file. Obviously, this wrapper is again a skeleton, but
608 you could mix and match this wrapper code with any of the code shown
609 above.
610
611 Alternately you could subclass the "HTML::Mason::ApacheHandler" class,
612 and override the "handler()" method it provides. See the Subclassing
613 documentation for more details. Of course, you could even create a
614 subclass and write a wrapper that called it.
615
617 This section describes how to set up common developer features.
618
619 Global Variables
620 Global variables can make programs harder to read, maintain, and debug,
621 and this is no less true for Mason components. Due to the persistent
622 mod_perl environment, globals require extra initialization and cleanup
623 care.
624
625 That said, there are times when it is very useful to make a value
626 available to all Mason components: a DBI database handle, a hash of
627 user session information, the server root for forming absolute URLs.
628
629 Because Mason by default parses components in "strict" mode, you'll
630 need to declare a global if you don't want to access it with an
631 explicit package name. The easiest way to declare a global is with the
632 allow_globals parameter.
633
634 Since all components run in the same package, you'll be able to set the
635 global in one component and access it in all the others.
636
637 Autohandlers are common places to assign values to globals. Use the
638 "<%once>" section if the global only needs to be initialized at load
639 time, or the "<%init>" section if it needs to be initialized every
640 request.
641
642 Sessions
643 Mason does not have a built-in session mechanism, but you can use the
644 "MasonX::Request::WithApacheSession" module, available from CPAN, to
645 add a session to every request. It can also automatically set and read
646 cookies containing the session id.
647
648 Data Caching
649 Data caching is implemented with DeWitt Clinton's "Cache::Cache"
650 module. For full understanding of this section you should read the
651 documentation for "Cache::Cache" as well as for relevant subclasses
652 (e.g. "Cache::FileCache").
653
654 Cache files
655 By default, "Cache::FileCache" is the subclass used for data
656 caching, although this may be overridden by the developer.
657 "Cache::FileCache" creates a separate subdirectory for every
658 component that uses caching, and one file some number of levels
659 underneath that subdirectory for each cached item. The root of the
660 cache tree is data_dir/"cache". The name of the cache subdirectory
661 for a component is determined by the function
662 "HTML::Mason::Utils::data_cache_namespace".
663
664 Default constructor options
665 Ordinarily, when "$m->cache" is called, Mason passes to the cache
666 constructor the "namespace", and "cache_root" options, along with
667 any other options given in the "$m->cache" method.
668
669 You may specify other default constructor options with the
670 data_cache_defaults parameter. For example,
671
672 PerlSetVar MasonDataCacheDefaults "cache_class => SizeAwareFileCache"
673 PerlAddVar MasonDataCacheDefaults "cache_depth => 2"
674 PerlAddVar MasonDataCacheDefaults "default_expires_in => 1 hour"
675
676 Any options passed to individual "$m->cache" calls override these
677 defaults.
678
679 Disabling data caching
680 If for some reason you want to disable data caching entirely, set
681 the default "cache_class" to "NullCache". This subclass faithfully
682 implements the cache API but never stores data.
683
685 This section explains Mason's various performance enhancements and how
686 to administer them. One of the best ways to maximize performance on
687 your production server is run in static_source mode; see the third
688 subsection below.
689
690 Code Cache
691 When Mason loads a component, it places it in a memory cache. By
692 default, the cache has no limit, but you can specify a maximum number
693 of components to cache with the code_cache_max_size parameter. In this
694 case, Mason will free up space as needed by discarding components. The
695 discard algorithm is least frequently used (LFU), with a periodic decay
696 to gradually eliminate old frequency information. In a nutshell, the
697 components called most often in recent history should remain in the
698 cache.
699
700 Previous versions of Mason attempted to estimate the size of each
701 component, but this proved so inaccurate as to be virtually useless for
702 cache policy. The max size is now specified purely in number of
703 components.
704
705 Mason can use certain optimizations with an unlimited cache, especially
706 in conjunction with static_source, so don't limit the cache unless
707 experience shows that your servers are growing too large. Many dynamic
708 sites can be served comfortably with all components in memory.
709
710 You can prepopulate the cache with components that you know will be
711 accessed often; see Preloading Components. Note that preloaded
712 components possess no special status in the cache and can be discarded
713 like any others.
714
715 Naturally, a cache entry is invalidated if the corresponding component
716 source file changes.
717
718 To turn off code caching completely, set code_cache_max_size to 0.
719
720 Object Files
721 The in-memory code cache is only useful on a per-process basis. Each
722 process must build and maintain its own cache. Shared memory caches are
723 conceivable in the future, but even those will not survive between web
724 server restarts.
725
726 As a secondary, longer-term cache mechanism, Mason stores a compiled
727 form of each component in an object file under data_dir/obj. Any server
728 process can eval the object file and save time on parsing the component
729 source file. The object file is recreated whenever the source file
730 changes.
731
732 The object file pathname is formed from three parts:
733
734 · the compiler "object_id" - this prevents different versions of
735 Mason or compilers from using the same object file, such as after
736 an upgrade
737
738 · the component path
739
740 · object_file_extension, by default ".obj"
741
742 Besides improving performance, object files can be useful for
743 debugging. If you feel the need to see what your source has been
744 translated into, you can peek inside an object file to see exactly how
745 Mason converted a given component to a Perl object. This was crucial
746 for pre-1.10 Mason, in which error line numbers were based on the
747 object file rather than the source file.
748
749 If for some reason you don't want Mason to create object files, set
750 use_object_files to 0.
751
752 Static Source Mode
753 In static_source mode, Mason assumes that the component hierarchy is
754 unchanging and thus does not check source timestamps when using an in-
755 memory cached component or object file. This significantly reduces
756 filesystem stats and other overhead. We've seen speedups by a factor of
757 two or three as a result of this mode, though of course YMMV.
758
759 When in static_source mode, you must remove object files and call
760 $interp->flush_code_cache in order for the server to recognize
761 component changes. The easiest way to arrange this is to point
762 static_source_touch_file to a file that can be touched whenever
763 components change.
764
765 We highly recommend running in this mode in production if you can
766 manage it. Many of Mason's future optimizations will be designed for
767 this mode. On development servers, of course, it makes sense to keep
768 this off so that components are reloaded automatically.
769
770 Disabling Autoflush
771 To support the dynamic autoflush feature, Mason has to check for
772 autoflush mode after printing every piece of text. If you can commit
773 to not using autoflush, setting enable_autoflush to 0 will allow Mason
774 to compile components more efficiently. Consider whether a few well-
775 placed "$m->flush_buffer" calls would be just as good as autoflush.
776
777 Write a handler subroutine
778 Writing your own "handler()" subroutine which uses an ApacheHandler
779 object (or objects) created during server startup is slightly faster
780 (around 5% or so) than configuring mason via your httpd.conf file and
781 letting Mason create its own ApacheHandler objects internally.
782
783 Preloading Components
784 You can tell Mason to preload a set of components in the parent
785 process, rather than loading them on demand, using the preloads
786 parameter. Each child server will start with those components loaded
787 in the memory cache. The trade-offs are:
788
789 time
790 a small one-time startup cost, but children save time by not having
791 to load the components
792
793 memory
794 a fatter initial server, but the memory for preloaded components
795 are shared by all children. This is similar to the advantage of
796 using modules only in the parent process.
797
798 Try to preload components that are used frequently and do not change
799 often. (If a preloaded component changes, all the children will have
800 to reload it from scratch.)
801
802 Preallocating the Output Buffer
803 You can set buffer_preallocate_size to set the size of the preallocated
804 output buffer for each request. This can reduce the number of
805 reallocations Perl performs as components output text.
806
808 When an error occurs, Mason can respond by:
809
810 · showing a detailed error message in the browser in HTML.
811
812 · die'ing, which sends a 500 status to the browser and lets the error
813 message go to the error logs.
814
815 The first behavior is ideal for development, where you want immediate
816 feedback on the error. The second behavior is usually desired for
817 production so that users are not exposed to messy error messages. You
818 choose the behavior by setting error_mode to "output" or "fatal"
819 respectively.
820
821 Error formatting is controlled by the error_format parameter. When
822 showing errors in the browser, Mason defaults to the "html" format.
823 When the error_mode is set to "fatal", the default format is "line",
824 which puts the entire error message on one line in a format suitable
825 for web server error logs. Mason also offers other formats, which are
826 covered in the Request class documentation.
827
828 Finally, you can use Apache's "ErrorDocument" directive to specify a
829 custom error handler for 500 errors. In this case, you'd set the
830 error_mode to "fatal". The URL specified by the "ErrorDocument"
831 directive could point to a Mason component.
832
833 Exceptions Under the Hood
834 The way that Mason really reports errors is through the use of
835 exception objects, which are implemented with the "Exception::Class"
836 module from CPAN, and some custom code in the HTML::Mason::Exceptions
837 module.
838
839 If, during the execution of a component, execution stops because some
840 code calls "die()", then Mason will catch this exception. If the
841 exception being thrown is just a string, then it will be converted to
842 an "HTML::Mason::Exception" object. If the exception being thrown is
843 an object with a "rethrow()" method, then this method will be called.
844 Otherwise, Mason simply leaves the exception untouched and calls
845 "die()" again.
846
847 Calling a Component to Handle Errors
848
849 Returning to the topic of wrapper code that we covered earlier, what if
850 you wanted to handle all request errors by calling an error handling
851 component? There is no way to do this without wrapper code. Here's an
852 example "handler()" subroutine that does this:
853
854 sub handler {
855 my ($r) = @_;
856
857 my $return = eval { $ah->handle_request($r) };
858
859 if ( my $err = $@ )
860 {
861 $r->pnotes( error => $err );
862 $r->filename( $r->document_root . '/error/500.html' );
863
864 return $ah->handle_request($r);
865 }
866
867 return $return;
868 }
869
870 First, we wrap our call to "$ah->handle_request()" in an "eval{}"
871 block. If an error occurs, we store it in the request object using the
872 "$r->pnotes()" method. Then we change the filename property of the
873 Apache request object to point to our error-handling component and call
874 the "$ah->handle_request()" method again, passing it the altered
875 request object. We could have put the exception in "$r->args", but we
876 want to leave this untouched so that the error-handling component can
877 see the original arguments.
878
879 Here's what that component error-handling component might look like:
880
881 <html>
882 <head>
883 <title>Error</title>
884 </head>
885
886 <body>
887
888 <p>
889 Looks like our application broke. Whatever you did, don't do it again!
890 </p>
891
892 <p>
893 If you have further questions, please feel free to contact us at <a
894 href="mailto:support@example.com">support@example.com</a>.
895 </p>
896
897 <p><a href="/">Click here</a> to continue.</p>
898
899 </body>
900 </html>
901
902 <%init>
903 my $error = $r->pnotes('error');
904
905 my $error_text = "Page is " . $r->parsed_uri->unparse . "\n\n";
906
907 $error_text .= UNIVERSAL::can( $error, 'as_text' ) ? $error->as_text : $error;
908
909 $r->log_error($error_text);
910
911 my $mail =
912 MIME::Lite->new
913 ( From => 'error-handler@example.com',
914 To => 'rt@example.com',
915 Subject => 'Application error',
916 Data => $error_text,
917 );
918
919 $r->register_cleanup( sub { $mail->send } );
920 </%init>
921
922 <%flags>
923 inherit => undef
924 </%flags>
925
926 This component does several things. First of all, it logs the complete
927 error to the Apache error logs, along with the complete URL, including
928 query string, that was requested. The "$r->parsed_uri()" method that
929 we use above is only available if the "Apache::URI" module has been
930 loaded.
931
932 The component also sends an email containing the error, in this case to
933 an RT installation, so that the error is logged in a bug tracking
934 system. Finally, it displays a less technical error message to the
935 user.
936
937 For this to work properly, you must set error_mode to "fatal", so that
938 Mason doesn't just display its own HTML error page.
939
941 Although Mason is most commonly used in conjunction with mod_perl, the
942 APIs are flexible enough to use in any environment. Below we describe
943 the two most common alternative environments, CGI and standalone
944 scripts.
945
946 Using Mason from a CGI Script
947 The easiest way to use Mason via a CGI script is with the CGIHandler
948 module module.
949
950 Here is a skeleton CGI script that calls a component and sends the
951 output to the browser.
952
953 #!/usr/bin/perl
954 use HTML::Mason::CGIHandler;
955
956 my $h = HTML::Mason::CGIHandler->new
957 (
958 data_dir => '/home/jethro/code/mason_data',
959 );
960
961 $h->handle_request;
962
963 The relevant portions of the httpd.conf file look like:
964
965 DocumentRoot /path/to/comp/root
966 ScriptAlias /cgi-bin/ /path/to/cgi-bin/
967
968 <LocationMatch "\.html$">
969 Action html-mason /cgi-bin/mason_handler.cgi
970 AddHandler html-mason .html
971 </LocationMatch>
972 <LocationMatch "^/cgi-bin/">
973 RemoveHandler .html
974 </LocationMatch>
975 <FilesMatch "(autohandler|dhandler)$">
976 Order allow,deny
977 Deny from all
978 </FilesMatch>
979
980 This simply causes Apache to call the mason_handler.cgi script every
981 time a URL ending in ".html" under the component root is requested.
982
983 To exclude certian directories from being under Mason control, you can
984 use something like the following:
985
986 <LocationMatch "^/(dir1|dir2|dir3)/">
987 RemoveHandler .html
988 </LocationMatch>
989
990 This script uses the CGIHandler class to do most of the heavy lifting.
991 See that class's documentation for more details.
992
993 Using Mason from a Standalone Script
994 Mason can be used as a pure text templating solution -- like
995 Text::Template and its brethren, but with more power (and of course
996 more complexity).
997
998 Here is a bare-bones script that calls a component file and sends the
999 result to standard output:
1000
1001 #!/usr/bin/perl
1002 use HTML::Mason;
1003 use strict;
1004
1005 my $interp = HTML::Mason::Interp->new ();
1006 $interp->exec(<relative path to file>, <args>...);
1007
1008 Because no component root was specified, the root is set to your
1009 current working directory. If you have a well defined and contained
1010 component tree, you'll probably want to specify a component root.
1011
1012 Because no data directory was specified, object files will not be
1013 created and data caching will not work in the default manner. If
1014 performance is an issue, you will want to specify a data directory.
1015
1016 Here's a slightly fuller script that specifies a component root and
1017 data directory, and captures the result in a variable rather than
1018 sending to standard output:
1019
1020 #!/usr/bin/perl
1021 use HTML::Mason;
1022 use strict;
1023
1024 my $outbuf;
1025 my $interp = HTML::Mason::Interp->new
1026 (comp_root => '/path/to/comp_root',
1027 data_dir => '/path/to/data_dir',
1028 out_method => \$outbuf
1029 );
1030 $interp->exec(<component-path>, <args>...);
1031
1032 # Do something with $outbuf
1033
1034
1035
1036perl v5.30.0 2019-07-26 HTML::Mason::Admin(3)