Bug #28584 for Catalyst-Plugin-Cache: Setting up Cache::FastMmap as backend fails

Tue Jul 31 08:54:01 2007 hl [...] muenchen-mail.de - Ticket created

Subject:	Setting up Cache::FastMmap as backend fails
Date:	Tue, 31 Jul 2007 14:53:46 +0200
To:	bug-Catalyst-Plugin-Cache [...] rt.cpan.org
From:	Herbert Leitz <hl [...] muenchen-mail.de>

Hi, this is not a really a bug, but an interoperability issue. I'm using Catalyst::Plugin::Cache-0.03 and Cache::FastMmap-1.16. If I specify Cache::FastMmap as a backend in myapp.yml: ... cache: backends: test: class: Cache::FastMmap expire_time: 1d share_file: /tmp/test_cache ... then setup_cache_backend_by_class() calls Cache::FastMmap->new(\%config) but it really needs an even-sized list Cache::FastMmap->new(%config) So maybe some provisions should be made for the different calling conventions of the backends. Herbert Leitz

Tue Jul 31 10:19:06 2007 NUFFIN [...] cpan.org - Taken

Tue Jul 31 10:34:40 2007 NUFFIN [...] cpan.org - Correspondence added

It's fixed in trunk, but I'm hesitent. Here's what I asked my army of Catalyst users to do: 17:30 <@nothingmuch> i need everyone to test C::P::Cache in trunk 17:31 <@nothingmuch> the change is when a backend is configured using class 17:31 <@nothingmuch> it'll first try to construct that class as $class->ew(%args); and then $class->new(\%args); 17:31 <@nothingmuch> it used to only do $class->new(%args); 17:31 <@nothingmuch> but typically passing \%args to something that then does my %args = @args will cause a warning, not an error 17:32 <@nothingmuch> wheras trying to deref the first hash key as a hash ref is an error, not a warning 17:32 <@nothingmuch> this is yucky 17:32 <@nothingmuch> but i think it's ever so slightly less yucky than keeping a list of which Cache backend like this and which likes that 17:32 <@nothingmuch> (FYI most of them accept both, some only \%argts, and Cache::FastMmap only %args); 17:32 <@nothingmuch> http://rt.cpan.org/Ticket/Display.html?id=28584 17:33 <@nothingmuch> another idea is to add a 'pass_ref' option 17:33 <@nothingmuch> so class => "Cache::FastMmap", pass_ref => 1, 17:33 <@nothingmuch> this risks no back compat problems at all 17:33 <@nothingmuch> but usability wise it's the last nice 17:33 <@nothingmuch> anyway, please report your findings minions! Your thoughts on the subject are more than welcome. Cheers,

Tue Jul 31 10:34:42 2007 The RT System itself - Status changed from 'new' to 'open'

Tue Jul 31 10:40:23 2007 NUFFIN [...] cpan.org - Correspondence added

For now a work around you could put the object there directly: $c->config->{cache}{backend} = Cache::FastMmap->new... Unfortunately this will not work with YAML

Tue Jul 31 14:35:31 2007 hl [...] muenchen-mail.de - Correspondence added

Subject:	Re: [rt.cpan.org #28584] Setting up Cache::FastMmap as backend fails
Date:	Tue, 31 Jul 2007 19:30:43 +0200
To:	bug-Catalyst-Plugin-Cache [...] rt.cpan.org
From:	Herbert Leitz <hl [...] muenchen-mail.de>

Am Dienstag, 31. Juli 2007 16:34 schrieb יובל קוג'מן via RT: Show quoted text

> <URL: http://rt.cpan.org/Ticket/Display.html?id=28584 > > > It's fixed in trunk, but I'm hesitent. > > Here's what I asked my army of Catalyst users to do: >

... Show quoted text

> > Your thoughts on the subject are more than welcome. > > Cheers,

Thanks for your fast response. Yes, this is a yucky problem. My suggestion is: use all three options. 1. Keep a list of well known modules. There aren't too many and their APIs will not change. 2. If a module is unknown, check 'pass_ref' 3. If pass_ref doesn't exist, try $class->new(%args) and then $class->new(\%args) This way, C::P::Cache will do the right thing and will work with as many different Modules as possible. Regards Herbert Leitz

Sat Jun 21 13:52:18 2008 NUFFIN [...] cpan.org - Correspondence added

Is this still an issue? I'm not up to date

Sun Jun 22 09:49:26 2008 Herbert.Leitz [...] t-online.de - Correspondence added

Subject:	Re: [rt.cpan.org #28584] Setting up Cache::FastMmap as backend fails
Date:	Sun, 22 Jun 2008 15:48:54 +0200
To:	bug-Catalyst-Plugin-Cache [...] rt.cpan.org
From:	Herbert Leitz <Herbert.Leitz [...] t-online.de>

Am Samstag, 21. Juni 2008 19:52:20 schrieb יובל קוג'מן via RT: Show quoted text

> <URL: http://rt.cpan.org/Ticket/Display.html?id=28584 > > > Is this still an issue? I'm not up to date

It has been fixed in Cache::FastMmap. Works fine now.

Sun Feb 08 12:19:38 2009 bobtfish [...] bobtfish.net - Status changed from 'open' to 'resolved'