On 7 Dec 2011, at 23:53, Charlie Garrison wrote:
I've got the opposite problem from that solved in v5.90007. I'm
running lighttpd/1.4.26 (on centos) so the LighttpdScriptNameFix
middleware should not be needed. But when I updated staging server,
our app broke, and applying LighttpdScriptNameFix unconditionally
fixed it (edited apply_default_middlewares in Catalyst.pm).
Bugger.
I don't understand SCRIPT_NAME and PATH_INO well enough to know what
*should* be happening. Where should I look to figure out why the
LighttpdScriptNameFix middleware still needs to be applied?
The relevant code is Catalyst::Engine::prepare_path
The most relevant tests are: t/aggregate/unit_core_engine-prepare_path.t
Any suggestions?
If you can patch the prepare_path method to grab a dump of $c->engine-
>env for a few paths (basically / and a path below /) for your cases,
and also the values for the output of $c->req->uri and $c->req->base
which are wrong, then let me know (.diff on that test best, but just
attaching a bunch of crap to a reply is also fine).
We'll see if we can work out how to get this working for everyone...
As a short term fix, however, instead of patching Catalyst, you're
probably better off making a myapp.psgi and applying the middleware in
that (as we won't stomp on it when you upgrade Catalyst)..
Cheers
t0m
_______________________________________________
List: [email protected]
Listinfo: http://lists.scsys.co.uk/cgi-bin/mailman/listinfo/catalyst
Searchable archive: http://www.mail-archive.com/[email protected]/
Dev site: http://dev.catalyst.perl.org/
