… | |
… | |
469 | |
469 | |
470 | Example: fork a process and wait for it |
470 | Example: fork a process and wait for it |
471 | |
471 | |
472 | my $done = AnyEvent->condvar; |
472 | my $done = AnyEvent->condvar; |
473 | |
473 | |
|
|
474 | # this forks and immediately calls exit in the child. this |
|
|
475 | # normally has all sorts of bad consequences for your parent, |
|
|
476 | # so take this as an example only. always fork and exec, |
|
|
477 | # or call POSIX::_exit, in real code. |
474 | my $pid = fork or exit 5; |
478 | my $pid = fork or exit 5; |
475 | |
479 | |
476 | my $w = AnyEvent->child ( |
480 | my $w = AnyEvent->child ( |
477 | pid => $pid, |
481 | pid => $pid, |
478 | cb => sub { |
482 | cb => sub { |
… | |
… | |
718 | This works because for every event source (EOF on file handle), |
722 | This works because for every event source (EOF on file handle), |
719 | there is one call to "begin", so the condvar waits for all calls to |
723 | there is one call to "begin", so the condvar waits for all calls to |
720 | "end" before sending. |
724 | "end" before sending. |
721 | |
725 | |
722 | The ping example mentioned above is slightly more complicated, as |
726 | The ping example mentioned above is slightly more complicated, as |
723 | the there are results to be passwd back, and the number of tasks |
727 | the there are results to be passed back, and the number of tasks |
724 | that are begun can potentially be zero: |
728 | that are begun can potentially be zero: |
725 | |
729 | |
726 | my $cv = AnyEvent->condvar; |
730 | my $cv = AnyEvent->condvar; |
727 | |
731 | |
728 | my %result; |
732 | my %result; |
… | |
… | |
1121 | Truly asynchronous (as opposed to non-blocking) I/O, should be in |
1125 | Truly asynchronous (as opposed to non-blocking) I/O, should be in |
1122 | the toolbox of every event programmer. AnyEvent::AIO transparently |
1126 | the toolbox of every event programmer. AnyEvent::AIO transparently |
1123 | fuses IO::AIO and AnyEvent together, giving AnyEvent access to |
1127 | fuses IO::AIO and AnyEvent together, giving AnyEvent access to |
1124 | event-based file I/O, and much more. |
1128 | event-based file I/O, and much more. |
1125 | |
1129 | |
|
|
1130 | AnyEvent::Fork, AnyEvent::Fork::RPC, AnyEvent::Fork::Pool, |
|
|
1131 | AnyEvent::Fork::Remote |
|
|
1132 | These let you safely fork new subprocesses, either locally or |
|
|
1133 | remotely (e.g.v ia ssh), using some RPC protocol or not, without the |
|
|
1134 | limitations normally imposed by fork (AnyEvent works fine for |
|
|
1135 | example). Dynamically-resized worker pools are obviously included as |
|
|
1136 | well. |
|
|
1137 | |
|
|
1138 | And they are quite tiny and fast as well - "abusing" AnyEvent::Fork |
|
|
1139 | just to exec external programs can easily beat using "fork" and |
|
|
1140 | "exec" (or even "system") in most programs. |
|
|
1141 | |
1126 | AnyEvent::Filesys::Notify |
1142 | AnyEvent::Filesys::Notify |
1127 | AnyEvent is good for non-blocking stuff, but it can't detect file or |
1143 | AnyEvent is good for non-blocking stuff, but it can't detect file or |
1128 | path changes (e.g. "watch this directory for new files", "watch this |
1144 | path changes (e.g. "watch this directory for new files", "watch this |
1129 | file for changes"). The AnyEvent::Filesys::Notify module promises to |
1145 | file for changes"). The AnyEvent::Filesys::Notify module promises to |
1130 | do just that in a portbale fashion, supporting inotify on GNU/Linux |
1146 | do just that in a portbale fashion, supporting inotify on GNU/Linux |
1131 | and some weird, without doubt broken, stuff on OS X to monitor |
1147 | and some weird, without doubt broken, stuff on OS X to monitor |
1132 | files. It can fall back to blocking scans at regular intervals |
1148 | files. It can fall back to blocking scans at regular intervals |
1133 | transparently on other platforms, so it's about as portable as it |
1149 | transparently on other platforms, so it's about as portable as it |
1134 | gets. |
1150 | gets. |
1135 | |
1151 | |
1136 | (I haven't used it myself, but I haven't heard anybody complaining |
1152 | (I haven't used it myself, but it seems the biggest problem with it |
1137 | about it yet). |
1153 | is it quite bad performance). |
1138 | |
1154 | |
1139 | AnyEvent::DBI |
1155 | AnyEvent::DBI |
1140 | Executes DBI requests asynchronously in a proxy process for you, |
1156 | Executes DBI requests asynchronously in a proxy process for you, |
1141 | notifying you in an event-based way when the operation is finished. |
1157 | notifying you in an event-based way when the operation is finished. |
1142 | |
|
|
1143 | AnyEvent::HTTPD |
|
|
1144 | A simple embedded webserver. |
|
|
1145 | |
1158 | |
1146 | AnyEvent::FastPing |
1159 | AnyEvent::FastPing |
1147 | The fastest ping in the west. |
1160 | The fastest ping in the west. |
1148 | |
1161 | |
1149 | Coro |
1162 | Coro |
… | |
… | |
2035 | usually happens when the first AnyEvent watcher is created, or the |
2048 | usually happens when the first AnyEvent watcher is created, or the |
2036 | library is loaded). |
2049 | library is loaded). |
2037 | |
2050 | |
2038 | If you have to fork, you must either do so *before* creating your first |
2051 | If you have to fork, you must either do so *before* creating your first |
2039 | watcher OR you must not use AnyEvent at all in the child OR you must do |
2052 | watcher OR you must not use AnyEvent at all in the child OR you must do |
2040 | something completely out of the scope of AnyEvent. |
2053 | something completely out of the scope of AnyEvent (see below). |
2041 | |
2054 | |
2042 | The problem of doing event processing in the parent *and* the child is |
2055 | The problem of doing event processing in the parent *and* the child is |
2043 | much more complicated: even for backends that *are* fork-aware or |
2056 | much more complicated: even for backends that *are* fork-aware or |
2044 | fork-safe, their behaviour is not usually what you want: fork clones all |
2057 | fork-safe, their behaviour is not usually what you want: fork clones all |
2045 | watchers, that means all timers, I/O watchers etc. are active in both |
2058 | watchers, that means all timers, I/O watchers etc. are active in both |
2046 | parent and child, which is almost never what you want. USing "exec" to |
2059 | parent and child, which is almost never what you want. Using "exec" to |
2047 | start worker children from some kind of manage rprocess is usually |
2060 | start worker children from some kind of manage prrocess is usually |
2048 | preferred, because it is much easier and cleaner, at the expense of |
2061 | preferred, because it is much easier and cleaner, at the expense of |
2049 | having to have another binary. |
2062 | having to have another binary. |
|
|
2063 | |
|
|
2064 | In addition to logical problems with fork, there are also implementation |
|
|
2065 | problems. For example, on POSIX systems, you cannot fork at all in Perl |
|
|
2066 | code if a thread (I am talking of pthreads here) was ever created in the |
|
|
2067 | process, and this is just the tip of the iceberg. In general, using fork |
|
|
2068 | from Perl is difficult, and attempting to use fork without an exec to |
|
|
2069 | implement some kind of parallel processing is almost certainly doomed. |
|
|
2070 | |
|
|
2071 | To safely fork and exec, you should use a module such as Proc::FastSpawn |
|
|
2072 | that let's you safely fork and exec new processes. |
|
|
2073 | |
|
|
2074 | If you want to do multiprocessing using processes, you can look at the |
|
|
2075 | AnyEvent::Fork module (and some related modules such as |
|
|
2076 | AnyEvent::Fork::RPC, AnyEvent::Fork::Pool and AnyEvent::Fork::Remote). |
|
|
2077 | This module allows you to safely create subprocesses without any |
|
|
2078 | limitations - you can use X11 toolkits or AnyEvent in the children |
|
|
2079 | created by AnyEvent::Fork safely and without any special precautions. |
2050 | |
2080 | |
2051 | SECURITY CONSIDERATIONS |
2081 | SECURITY CONSIDERATIONS |
2052 | AnyEvent can be forced to load any event model via |
2082 | AnyEvent can be forced to load any event model via |
2053 | $ENV{PERL_ANYEVENT_MODEL}. While this cannot (to my knowledge) be used |
2083 | $ENV{PERL_ANYEVENT_MODEL}. While this cannot (to my knowledge) be used |
2054 | to execute arbitrary code or directly gain access, it can easily be used |
2084 | to execute arbitrary code or directly gain access, it can easily be used |