| Acme-State documentation | Contained in the Acme-State distribution. |
Acme::State - Save application state on exit and restores state on startup
use Acme::State;
our $t;
print "t: $t\n";
$t = int rand 100;
print "new t: $t\n";
... and then run it again.
Crawls the package hierarchy looking for our variables.
Stores them all off in a file in the home directory of the user running the script.
When the script using this module starts up, this same file is read in and the
variables are restored.
Serializes scalars, hashes, and arrays declared using our, use vars, or otherwise
not declared using my.
Uses Storable to write the data.
The save is placed in the home directory of the user the script is executing as.
The file name is the same as the script's name ($0) plus ".save".
It also keeps one backup around, named $0.save.last, and it may leave a
$0.save.new if interrupted.
Web apps written using Continuity get persistant state, so why shouldn't command line apps? Hey, and maybe Continuity apps want to persist some state in case the server implodes. Who knows.
$Acme::State::wantcoderefs, if set true, takes things a step further and tells
Acme::State to also serialize subroutines it finds.
Nothing says fun like persisting coderefs from the stash and a 40 of Mickey's.
This code reserves the right to die if anything goes horribly wrong.
Explicitly request a snapshot of state be written to disc.
dies if unable to write the save file or if a sanity check fails.
Optionally also use Coro to create an execution context that runs peroidically to save snapshots.
Original version; created by h2xs 1.23 with options
-A -C -X -b 5.8.0 -c -n Stupid::State
PAUSE rejected the first one because it didn't like the permissions h2xs left for the automatically generated META.yml file so it wouldn't index it, but it also wouldn't let me delete it, so this version is actually identical to 0.01.
Ooops, actually use IO::Handle. Not every program already does that for us.
What could possibily go wrong?
You *could* use an ORM, and wind up translating all of your data to a relational schema you
don't care about or else have it automatically mapped and completely miss the point of
using a relational database.
You *could* just store your data in the Ether with Memcached.
You could INSERT and UPDATE manually against a database to store every little tidbit and factoid
as they're computed.
You could use BerekelyDB, including the build-in legacy dbmopen and mangle everything
down to a flat associative list.
You could use Data::Dumper to write a structure to a file and eval that on startup
and keep all of your precious application data in one big datastructure and still not be able to
persist entire objects.
You could use dump and keep waiting for the day that someone finally writes undump.
But what's the fun in that?
None of those are one use line and then never another thought.
That's like work for something.
Work is for suckers.
We're Perl programmers.
If it's not automatic, it's not worth doing.
Scott Walters, <scott@slowass.net>
Copyright (C) 2009 by Scott Walters
This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself, either Perl version 5.8.0 or, at your option, any later version of Perl 5 you may have available.
| Acme-State documentation | Contained in the Acme-State distribution. |
package Acme::State; use 5.008000; use strict; use warnings; our $VERSION = '0.03'; use B; use Storable; use Devel::Caller 'caller_cv'; use IO::Handle; my @stop_modules = ( '1' .. '9', ':', 'SIG', 'stderr', '__ANON__', 'utf8::', 'CORE::', 'DynaLoader::', 'strict::', 'stdout', 'attributes::', 'stdin', 'ARGV', 'INC', 'Scalar::', 'ENV', 'Regexp::', 'XSLoader::', 'UNIVERSAL::', 'overload::', 'B::', 'Carp::', 'Data::', 'PerlIO::', '0', 'BEGIN', 'STDOUT', 'IO::', '_', 'Dumper', 'Exporter::', 'bytes::', 'STDERR', 'Internals::', 'STDIN', 'Config::', 'warnings::', 'DB::', 'APR::', 'Apache2::', 'Apache::', 'autobox::', 'BSD::', 'CGITempFile::', 'Compress::', 'Devel::', 'Dos::', 'EPOC::', 'Encode::', 'Fh::', 'File::', 'HTTP::', 'LWP::', 'List::', 'Log::', 'MIME::', 'Mac::', 'MacPerl::', 'O::', 'POSIX::', 'Scope::', 'Sys::', 'Term::', 'Thread::', 'Time::', 'VMS::', 'fields::', 'blackhole::', 'Autobox::', 'Module::', 'Win32::', 'MultipartBuffer::', 'q::', 'sort::', ); sub import { my $save_fn = save_file_name(); if(-f $save_fn) { local $Storable::Eval = 1; my $save = Storable::retrieve $save_fn; sub { my $package = shift; my $tree = shift; no strict 'refs'; for my $k (keys %$tree) { if($k =~ m/::$/) { caller_cv(0)->($package.$k, $tree->{$k}); } elsif(ref($tree->{$k})) { *{$package.$k} = $tree->{$k}; } else { die $package.$k . " doesn't contain a ref"; } } }->('main::', $save); } } sub save_file_name { my $zero = $0 || 'untitledprogram'; $zero =~ s{.*/}{}; return +(getpwuid $<)[7].'/'.$zero.'.store'; } sub save_state { our $wantcoderefs; my $tree = sub { my $package = shift; my $node = shift() || { }; no strict 'refs'; for my $k (keys %$package) { next if $k =~ m/main::$/; next if $k =~ m/[^\w:]/; next if grep $_ eq $k, @stop_modules; if($k =~ m/::$/) { # recurse into that namespace unless it corresponds to a .pm module that got used at some point my $modulepath = $package.$k; for($modulepath) { s{^main::}{}; s{::$}{}; s{::}{/}g; $_ .= '.pm'; } next if exists $INC{$modulepath}; $node->{$k} ||= { }; caller_cv(0)->($package.$k, $node->{$k}); } elsif( *{$package.$k}{HASH} ) { $node->{$k} = *{$package.$k}{HASH}; } elsif( *{$package.$k}{ARRAY} ) { $node->{$k} = *{$package.$k}{ARRAY}; } elsif( *{$package.$k}{CODE} ) { next unless $wantcoderefs; # save coderefs but only if they aren't XS (can't serialize those) and weren't exported from elsewhere. my $ob = B::svref_2object(*{$package . $k}{CODE}); my $rootop = $ob->ROOT; my $stashname = $$rootop ? $ob->STASH->NAME . '::' : '(none)'; if($$rootop and ($stashname eq $package or 'main::'.$stashname eq $package or $stashname eq 'main::' )) { # when we eval something in code in main::, it comes up as being exported from main::. *sigh* $node->{$k} = *{$package . $k}{CODE}; } } else { $node->{$k} = *{$package.$k}{SCALAR} unless ref(*{$package.$k}{SCALAR}) eq 'GLOB'; } } return $node; }->('main::'); # use Data::Dumper; print "debug: ", Data::Dumper::Dumper($tree), "\n"; local $Storable::Deparse = $wantcoderefs; my $save_fn = save_file_name(); # $save_fn =~ s{/-}{/x}g; warn "saving to: ``$save_fn.new''"; Storable::nstore $tree, $save_fn.'.new' or die "saving state failed: $!"; # warn "okay, Storable::nstore done"; rename $save_fn, $save_fn.'.last'; # it's okay if it fails... file might not exist rename $save_fn.'.new', $save_fn or die "renaming new save file into place as ``$save_fn'' failed: $!"; return 1; } END { STDERR->print("Acme::State: Saving program state!\n\n"); save_state(); };
__END__ scraps... async { my $timer = Coro::Event->timer( interval => 2, ); my $last_save_time = time; # my $mod_time = -M __FILE__; while(1) { $timer->next; if(time - $last_save_time > 60*15) { $save_db->(); $last_save_time = time; } } }; # if(-M __FILE__ != $mod_time) { # $save_db->(); # STDERR->print("Exec-ing self!\n\n"); # system '/usr/bin/perl', '-c', __FILE__ and do { $mod_time = -M __FILE__; next; }; # exec '/usr/bin/perl', __FILE__; # } # STDERR->print("deteced code imported from elsewhere: ``$package$k'' was imported from ``$stashname''\n") if $$rootop and ($stashname ne $package and 'main::'.$stashname ne $package and $stashname ne 'main::');