[ViewVC] Annotation of: cvs/Guard/README

NAME
    Guard - safe cleanup blocks

SYNOPSIS
       use Guard;
   
       # temporarily chdir to "/etc" directory, but make sure
       # to go back to "/" no matter how myfun exits:
       sub myfun {
          scope_guard { chdir "/" };
          chdir "/etc";
   
          code_that_might_die_or_does_other_fun_stuff;
       }

       # create an object that, when the last reference to it is gone,
       # invokes the given codeblock:
       my $guard = guard { print "destroyed!\n" };
       undef $guard; # probably destroyed here

DESCRIPTION
    This module implements so-called "guards". A guard is something (usually
    an object) that "guards" a resource, ensuring that it is cleaned up when
    expected.

    Specifically, this module supports two different types of guards: guard
    objects, which execute a given code block when destroyed, and scoped
    guards, which are tied to the scope exit.

FUNCTIONS
    This module currently exports the "scope_guard" and "guard" functions by
    default.

    scope_guard BLOCK
        Registers a block that is executed when the current scope (block,
        function, method, eval etc.) is exited.

        See the EXCEPTIONS section for an explanation of how exceptions
        (i.e. "die") are handled inside guard blocks.

        The description below sounds a bit complicated, but that's just
        because "scope_guard" tries to get even corner cases "right": the
        goal is to provide you with a rock solid clean up tool.

        The behaviour is similar to this code fragment:

           eval ... code following scope_guard ...
           {
              local $@;
              eval BLOCK;
              eval { $Guard::DIED->() } if $@;
           }
           die if $@;

        Except it is much faster, and the whole thing gets executed even
        when the BLOCK calls "exit", "goto", "last" or escapes via other
        means.

        If multiple BLOCKs are registered to the same scope, they will be
        executed in reverse order. Other scope-related things such as
        "local" are managed via the same mechanism, so variables "local"ised
        *after* calling "scope_guard" will be restored when the guard runs.

        Example: temporarily change the timezone for the current process,
        ensuring it will be reset when the "if" scope is exited:

           use Guard;
           use POSIX ();

           if ($need_to_switch_tz) {
              # make sure we call tzset after $ENV{TZ} has been restored
              scope_guard { POSIX::tzset };

              # localise after the scope_guard, so it gets undone in time
              local $ENV{TZ} = "Europe/London";
              POSIX::tzset;

              # do something with the new timezone
           }

    my $guard = guard BLOCK
        Behaves the same as "scope_guard", except that instead of executing
        the block on scope exit, it returns an object whose lifetime
        determines when the BLOCK gets executed: when the last reference to
        the object gets destroyed, the BLOCK gets executed as with
        "scope_guard".

        See the EXCEPTIONS section for an explanation of how exceptions
        (i.e. "die") are handled inside guard blocks.

        Example: acquire a Coro::Semaphore for a second by registering a
        timer. The timer callback references the guard used to unlock it
        again. (Please ignore the fact that "Coro::Semaphore" has a "guard"
        method that does this already):

           use Guard;
           use Coro::AnyEvent;
           use Coro::Semaphore;

           my $sem = new Coro::Semaphore;

           sub lock_for_a_second {
              $sem->down;
              my $guard = guard { $sem->up };

              Coro::AnyEvent::sleep 1;

              # $sem->up gets executed when returning
           }

        The advantage of doing this with a guard instead of simply calling
        "$sem->down" in the callback is that you can opt not to create the
        timer, or your code can throw an exception before it can create the
        timer (or the thread gets canceled), or you can create multiple
        timers or other event watchers and only when the last one gets
        executed will the lock be unlocked. Using the "guard", you do not
        have to worry about catching all the places where you have to unlock
        the semaphore.

    $guard->cancel
        Calling this function will "disable" the guard object returned by
        the "guard" function, i.e. it will free the BLOCK originally passed
        to "guard "and will arrange for the BLOCK not to be executed.

        This can be useful when you use "guard" to create a cleanup handler
        to be called under fatal conditions and later decide it is no longer
        needed.

EXCEPTIONS
    Guard blocks should not normally throw exceptions (that is, "die").
    After all, they are usually used to clean up after such exceptions.
    However, if something truly exceptional is happening, a guard block
    should of course be allowed to die. Also, programming errors are a large
    source of exceptions, and the programmer certainly wants to know about
    those.

    Since in most cases, the block executing when the guard gets executed
    does not know or does not care about the guard blocks, it makes little
    sense to let containing code handle the exception.

    Therefore, whenever a guard block throws an exception, it will be caught
    by Guard, followed by calling the code reference stored in $Guard::DIED
    (with $@ set to the actual exception), which is similar to how most
    event loops handle this case.

    The default for $Guard::DIED is to call "warn "$@"", i.e. the error is
    printed as a warning and the program continues.

    The $@ variable will be restored to its value before the guard call in
    all cases, so guards will not disturb $@ in any way.

    The code reference stored in $Guard::DIED should not die (behaviour is
    not guaranteed, but right now, the exception will simply be ignored).

AUTHOR
     Marc Lehmann <schmorp@schmorp.de>
     http://home.schmorp.de/

THANKS
    Thanks to Marco Maisenhelder, who reminded me of the $Guard::DIED
    solution to the problem of exceptions.

SEE ALSO
    Scope::Guard and Sub::ScopeFinalizer, which actually implement dynamic
    guards only, not scoped guards, and have a lot higher CPU, memory and
    typing overhead.

    Hook::Scope, which has apparently never been finished and can corrupt
    memory when used.

Revision:	1.6
Committed:	Fri Mar 12 17:25:58 2010 UTC (14 years, 2 months ago) by root
Branch:	MAIN
Changes since 1.5:	+23 -24 lines
Log Message:	* empty log message *
#	User	Rev	Content
1	root	1.1	NAME
2	root	1.2	Guard - safe cleanup blocks
3	root	1.1
4			SYNOPSIS
5	root	1.3	use Guard;
6	root	1.5
7			# temporarily chdir to "/etc" directory, but make sure
8	root	1.3	# to go back to "/" no matter how myfun exits:
9			sub myfun {
10			scope_guard { chdir "/" };
11			chdir "/etc";
12	root	1.5
13			code_that_might_die_or_does_other_fun_stuff;
14	root	1.3	}
15	root	1.1
16	root	1.5	# create an object that, when the last reference to it is gone,
17			# invokes the given codeblock:
18			my $guard = guard { print "destroyed!\n" };
19			undef $guard; # probably destroyed here
20
21	root	1.1	DESCRIPTION
22	root	1.2	This module implements so-called "guards". A guard is something (usually
23			an object) that "guards" a resource, ensuring that it is cleaned up when
24			expected.
25
26			Specifically, this module supports two different types of guards: guard
27			objects, which execute a given code block when destroyed, and scoped
28			guards, which are tied to the scope exit.
29
30			FUNCTIONS
31			This module currently exports the "scope_guard" and "guard" functions by
32			default.
33
34			scope_guard BLOCK
35			Registers a block that is executed when the current scope (block,
36			function, method, eval etc.) is exited.
37
38	root	1.3	See the EXCEPTIONS section for an explanation of how exceptions
39			(i.e. "die") are handled inside guard blocks.
40
41	root	1.2	The description below sounds a bit complicated, but that's just
42			because "scope_guard" tries to get even corner cases "right": the
43			goal is to provide you with a rock solid clean up tool.
44
45	root	1.3	The behaviour is similar to this code fragment:
46	root	1.2
47			eval ... code following scope_guard ...
48			{
49			local $@;
50			eval BLOCK;
51			eval { $Guard::DIED->() } if $@;
52			}
53			die if $@;
54
55			Except it is much faster, and the whole thing gets executed even
56			when the BLOCK calls "exit", "goto", "last" or escapes via other
57			means.
58
59	root	1.3	If multiple BLOCKs are registered to the same scope, they will be
60			executed in reverse order. Other scope-related things such as
61			"local" are managed via the same mechanism, so variables "local"ised
62			after calling "scope_guard" will be restored when the guard runs.
63
64			Example: temporarily change the timezone for the current process,
65			ensuring it will be reset when the "if" scope is exited:
66
67			use Guard;
68			use POSIX ();
69
70			if ($need_to_switch_tz) {
71			# make sure we call tzset after $ENV{TZ} has been restored
72			scope_guard { POSIX::tzset };
73
74			# localise after the scope_guard, so it gets undone in time
75			local $ENV{TZ} = "Europe/London";
76			POSIX::tzset;
77	root	1.2
78	root	1.3	# do something with the new timezone
79	root	1.2	}
80
81			my $guard = guard BLOCK
82			Behaves the same as "scope_guard", except that instead of executing
83			the block on scope exit, it returns an object whose lifetime
84			determines when the BLOCK gets executed: when the last reference to
85			the object gets destroyed, the BLOCK gets executed as with
86			"scope_guard".
87
88	root	1.3	See the EXCEPTIONS section for an explanation of how exceptions
89			(i.e. "die") are handled inside guard blocks.
90	root	1.2
91			Example: acquire a Coro::Semaphore for a second by registering a
92			timer. The timer callback references the guard used to unlock it
93	root	1.3	again. (Please ignore the fact that "Coro::Semaphore" has a "guard"
94			method that does this already):
95	root	1.2
96	root	1.3	use Guard;
97	root	1.6	use Coro::AnyEvent;
98	root	1.2	use Coro::Semaphore;
99
100			my $sem = new Coro::Semaphore;
101
102	root	1.3	sub lock_for_a_second {
103	root	1.2	$sem->down;
104			my $guard = guard { $sem->up };
105
106	root	1.6	Coro::AnyEvent::sleep 1;
107
108			# $sem->up gets executed when returning
109	root	1.2	}
110
111			The advantage of doing this with a guard instead of simply calling
112			"$sem->down" in the callback is that you can opt not to create the
113			timer, or your code can throw an exception before it can create the
114	root	1.6	timer (or the thread gets canceled), or you can create multiple
115			timers or other event watchers and only when the last one gets
116			executed will the lock be unlocked. Using the "guard", you do not
117			have to worry about catching all the places where you have to unlock
118			the semaphore.
119	root	1.2
120	root	1.3	$guard->cancel
121	root	1.2	Calling this function will "disable" the guard object returned by
122			the "guard" function, i.e. it will free the BLOCK originally passed
123			to "guard "and will arrange for the BLOCK not to be executed.
124
125	root	1.6	This can be useful when you use "guard" to create a cleanup handler
126			to be called under fatal conditions and later decide it is no longer
127			needed.
128	root	1.2
129			EXCEPTIONS
130	root	1.3	Guard blocks should not normally throw exceptions (that is, "die").
131			After all, they are usually used to clean up after such exceptions.
132			However, if something truly exceptional is happening, a guard block
133	root	1.6	should of course be allowed to die. Also, programming errors are a large
134			source of exceptions, and the programmer certainly wants to know about
135			those.
136	root	1.2
137	root	1.4	Since in most cases, the block executing when the guard gets executed
138	root	1.2	does not know or does not care about the guard blocks, it makes little
139			sense to let containing code handle the exception.
140
141	root	1.6	Therefore, whenever a guard block throws an exception, it will be caught
142			by Guard, followed by calling the code reference stored in $Guard::DIED
143	root	1.4	(with $@ set to the actual exception), which is similar to how most
144			event loops handle this case.
145	root	1.2
146	root	1.6	The default for $Guard::DIED is to call "warn "$@"", i.e. the error is
147			printed as a warning and the program continues.
148	root	1.3
149			The $@ variable will be restored to its value before the guard call in
150			all cases, so guards will not disturb $@ in any way.
151
152	root	1.2	The code reference stored in $Guard::DIED should not die (behaviour is
153			not guaranteed, but right now, the exception will simply be ignored).
154
155	root	1.1	AUTHOR
156			Marc Lehmann <schmorp@schmorp.de>
157			http://home.schmorp.de/
158
159	root	1.2	THANKS
160	root	1.3	Thanks to Marco Maisenhelder, who reminded me of the $Guard::DIED
161			solution to the problem of exceptions.
162	root	1.2
163	root	1.4	SEE ALSO
164	root	1.6	Scope::Guard and Sub::ScopeFinalizer, which actually implement dynamic
165			guards only, not scoped guards, and have a lot higher CPU, memory and
166			typing overhead.
167	root	1.4
168	root	1.6	Hook::Scope, which has apparently never been finished and can corrupt
169	root	1.4	memory when used.
170