Florian Ragwitz [Wed, 2 Mar 2011 11:46:17 +0000]
lighttpd and iis6 now work
Florian Ragwitz [Wed, 2 Mar 2011 11:38:47 +0000]
Reduce code repetition
Florian Ragwitz [Wed, 2 Mar 2011 11:32:22 +0000]
Add legacy workaround for nginx PATH_INFO
Florian Ragwitz [Wed, 2 Mar 2011 11:19:12 +0000]
Make legacy IIS env fixing work again
Florian Ragwitz [Wed, 2 Mar 2011 11:13:41 +0000]
Make legacy lighttpd env fixing work again
Florian Ragwitz [Wed, 2 Mar 2011 11:13:24 +0000]
Load middlewares where they're actually used
Florian Ragwitz [Wed, 2 Mar 2011 10:57:39 +0000]
Make _wrapped_legacy_psgi_app do wrapping only
We now pass in the psgi app to wrap, which should make it quite a bit easier to
test. We'll need that to check if the various quirk fixes for different web
servers actually work.
Florian Ragwitz [Wed, 2 Mar 2011 10:30:51 +0000]
Version 5.89002
Florian Ragwitz [Wed, 2 Mar 2011 10:02:51 +0000]
The HTTP engine is no more, and was also never required for this test
Florian Ragwitz [Wed, 2 Mar 2011 09:59:19 +0000]
Avoid loading actual engines in script tests
This way we can run these tests without having FCGI.pm installed.
Florian Ragwitz [Tue, 1 Mar 2011 18:37:44 +0000]
PSGI, not Plack
Florian Ragwitz [Tue, 1 Mar 2011 18:37:23 +0000]
Update Upgrading docs
Florian Ragwitz [Tue, 1 Mar 2011 18:30:31 +0000]
Make getting to a PSGI app saner
All the user should ever need to call is ->psgi_app (previously raw_psgi_app),
to get an unwrapped psgi app as provided by the engine. This is what should be
used when writing .psgi files.
For all other ways of running catalyst apps there's no need to be able to get to
the psgi application that'll actually run, so here we're making everything
that's responsible for reading .psgi files to run them with Catalyst::Script::*
or for wrapping the plain psgi code ref with back-compat middleware private.
Unfortunately this removes some of the docs about the automatic .psgi
loading. These should be added again elsewhere. Also the Upgrading docs need
updating after this.
Florian Ragwitz [Tue, 1 Mar 2011 14:26:46 +0000]
Version 5.89001
Florian Ragwitz [Tue, 1 Mar 2011 14:38:40 +0000]
Stop psgi_file.t from recursing
psgi_app is *not* to be used in .psgi files. It's the compat layer that'll give
you either an app loaded from from a .psgi file, or the default PSGI app with
the reverse proxy middleware and stuff.
In .psgi files only raw_psgi_app should ever be used, as that'll return the bare
catalyst psgi app to then be wrapped in middlewares manually.
I see how the naming is shit and confusing though. Suggestions welcome.
Florian Ragwitz [Tue, 1 Mar 2011 14:22:10 +0000]
Rename Engine::Loader to EngineLoader
It's not an engine, as those exist no more, so move it out of the ::Engine::
namespace.
Florian Ragwitz [Tue, 1 Mar 2011 14:20:08 +0000]
Changelog updates
Florian Ragwitz [Tue, 1 Mar 2011 13:49:57 +0000]
Merge branch 'master' into psgi
* master: (22 commits)
Change repos metadata to git
Version 5.80032
Pass the extra restart options
Fix test
Changelog
added myself (dd070) in contributors list
removed duplicate parameters in return value of _restarter_args
added more parameters in return value of _restarter_args
Add a few tests
Patch to make restarter class configurable / settable in prefs
Apply more correct fix than 8df53b (I hope)
Rename test
Make tests more clear
Merge revert
8df53bed
Un-TODO abraxxa's tests
This scares me, but it fixes stuf work work
Fix 5.80 bug which causes slurp to fail if called multiple times
Fix body predicate bug/back compat issue
Version 5.80031
added test for chained dispatcher fail on multiple CaptureArgs(0) parts introduced in 5.80030
...
Conflicts:
Changes
Makefile.PL
lib/Catalyst.pm
lib/Catalyst/Engine/FastCGI.pm
lib/Catalyst/Runtime.pm
Florian Ragwitz [Tue, 1 Mar 2011 13:43:05 +0000]
Properly skip known broken tests
Faked passing tests are not cool, and nor are __END__ sections, as those will
break aggregation.
If we actually want to prevent the body of the tests from being compiled,
putting the C<< plan skip_all => ... >> in a BEGIN block is the right thing to
do.
Tomas Doran [Mon, 28 Feb 2011 19:41:54 +0000]
Change repos metadata to git
Tomas Doran [Wed, 23 Feb 2011 01:10:30 +0000]
Version 5.80032
Tomas Doran [Wed, 23 Feb 2011 00:47:16 +0000]
Pass the extra restart options
Tomas Doran [Wed, 23 Feb 2011 00:46:13 +0000]
Fix test
Tomas Doran [Tue, 22 Feb 2011 22:19:16 +0000]
Changelog
Tomas Doran [Tue, 22 Feb 2011 22:10:31 +0000]
Merge branch 'restarter_changeclass' into 5.80032_dev
* restarter_changeclass:
added myself (dd070) in contributors list
removed duplicate parameters in return value of _restarter_args
added more parameters in return value of _restarter_args
Add a few tests
Patch to make restarter class configurable / settable in prefs
Dhaval Dhanani [Wed, 16 Feb 2011 16:04:55 +0000]
added myself (dd070) in contributors list
Dhaval Dhanani [Wed, 16 Feb 2011 15:40:29 +0000]
removed duplicate parameters in return value of _restarter_args
Dhaval Dhanani [Wed, 16 Feb 2011 15:35:10 +0000]
added more parameters in return value of _restarter_args
Tomas Doran [Mon, 14 Feb 2011 21:12:25 +0000]
Add a few tests
Tomas Doran [Mon, 14 Feb 2011 21:05:30 +0000]
Patch to make restarter class configurable / settable in prefs
Tomas Doran [Tue, 22 Feb 2011 22:09:40 +0000]
Merge branch 'chained_fix' into 5.80032_dev
* chained_fix:
Apply more correct fix than 8df53b (I hope)
Rename test
Make tests more clear
Merge revert
8df53bed
Un-TODO abraxxa's tests
Tomas Doran [Tue, 22 Feb 2011 01:31:11 +0000]
Apply more correct fix than 8df53b (I hope)
Tomas Doran [Tue, 22 Feb 2011 01:30:49 +0000]
Rename test
Tomas Doran [Tue, 22 Feb 2011 01:15:29 +0000]
Make tests more clear
Tomas Doran [Wed, 16 Feb 2011 21:43:06 +0000]
Tomas Doran [Wed, 16 Feb 2011 21:33:35 +0000]
Un-TODO abraxxa's tests
Tomas Doran [Tue, 22 Feb 2011 21:43:07 +0000]
This scares me, but it fixes stuf work work
Tomas Doran [Tue, 15 Feb 2011 16:10:03 +0000]
Fix 5.80 bug which causes slurp to fail if called multiple times
Tomas Doran [Mon, 14 Feb 2011 21:35:11 +0000]
Fix body predicate bug/back compat issue
Tomas Doran [Fri, 4 Feb 2011 00:48:29 +0000]
Dink docs to be more clear on psgi files / converting
Tomas Doran [Fri, 4 Feb 2011 00:38:29 +0000]
Recommending infinite loops in the docs is bad
Tomas Doran [Tue, 1 Feb 2011 21:50:48 +0000]
Update changelog and upgrading
Tomas Doran [Tue, 1 Feb 2011 21:50:10 +0000]
Add failing test of using a psgi file. Are you only meant to call ->raw_psgi_app, or?
Tomas Doran [Tue, 1 Feb 2011 21:49:29 +0000]
Delete obslolete crap from ROADMAP, update version numbers
Tomas Doran [Mon, 31 Jan 2011 09:29:25 +0000]
Tweaks
Tomas Doran [Mon, 31 Jan 2011 09:16:11 +0000]
Fix Engine::Stomp with psgi
Tomas Doran [Mon, 31 Jan 2011 08:24:44 +0000]
Version 5.80031
Alexander Hartmaier [Fri, 28 Jan 2011 14:09:29 +0000]
added test for chained dispatcher fail on multiple CaptureArgs(0) parts introduced in 5.80030
Tomas Doran [Thu, 27 Jan 2011 11:01:41 +0000]
Fix undef warning in Engine::FastCGI
John Napiorkowski [Tue, 25 Jan 2011 14:50:11 +0000]
added Engine::PSGI instructions
Darren Duncan [Tue, 25 Jan 2011 00:42:39 +0000]
fix =head levels for 'Upgrading to Catalyst 5.90'
Tomas Doran [Mon, 24 Jan 2011 23:31:45 +0000]
Don't be dumb, auth breaks on new cat, not other way round
Tomas Doran [Mon, 24 Jan 2011 23:28:10 +0000]
Small docs and fixes
Andrew Rodland [Mon, 24 Jan 2011 19:57:47 +0000]
Beat the args handling in Script::FastCGI into working condition.
Tomas Doran [Mon, 24 Jan 2011 11:29:24 +0000]
Bump dep
Florian Ragwitz [Mon, 24 Jan 2011 10:50:18 +0000]
Version 5.89000
This is a trial release from the psgi branch.
Florian Ragwitz [Sun, 23 Jan 2011 15:42:37 +0000]
Merge branch 'master' into psgi
master:
Version 5.80030
Fix a warning with undef bodies
Changelog the Package::Stash bugfix
Changelog the resp->body improvements
Work with PP Package::Stash again now the incorrect assumption that broke with ::XS is fixed
Stop relying on Package::Stash PP bugs
Add conflict
Bad test failed to show bug.
Fix the case for body '0'
Make response body able to be undef to allow RenderView to see 'defined but empty' body, allowing X-Sendfile to work nicer. Also removes horrible modifier code. win/win if it doesn't break anything else.
Extra links in docs
Add test case for uri_for() with #fragment and query params, broken by 5.7008 when uri_for() was reimplemented without URI.pm.
Found a another fault in chained action dispatcher.
Conflicts:
lib/Catalyst/Engine/CGI.pm
Tomas Doran [Thu, 20 Jan 2011 00:01:44 +0000]
Hit the docs some
Tomas Doran [Wed, 19 Jan 2011 23:44:06 +0000]
Additional notes and cleanup
Florian Ragwitz [Tue, 4 Jan 2011 12:55:54 +0000]
Version 5.80030
Florian Ragwitz [Tue, 4 Jan 2011 12:55:50 +0000]
Fix a warning with undef bodies
Florian Ragwitz [Tue, 4 Jan 2011 12:12:20 +0000]
Changelog the Package::Stash bugfix
Florian Ragwitz [Tue, 4 Jan 2011 12:12:16 +0000]
Changelog the resp->body improvements
John Napiorkowski [Tue, 4 Jan 2011 04:03:39 +0000]
first go at upgrading docs
Florian Ragwitz [Mon, 3 Jan 2011 20:20:52 +0000]
Work with PP Package::Stash again now the incorrect assumption that broke with ::XS is fixed
Florian Ragwitz [Mon, 3 Jan 2011 19:59:37 +0000]
Stop relying on Package::Stash PP bugs
We want to know if $MyApp::_config contains something like a valid
configuration, not if the package symbol exists.
The pure-perl version Package::Stash has always been buggy in assuming that a
scalar symbol containing undef is a symbol that doesn't exist. This is to work
around the fact that getting the SCALAR slot from a glob in pure perl
autovivifies the scalar to undef. Package::Stash::XS doesn't have that
limitation and is used transparently by Package::Stash if it's installed. That
exposed this bug in our code.
Tomas Doran [Sat, 25 Dec 2010 14:24:15 +0000]
Add conflict
Tomas Doran [Fri, 17 Dec 2010 00:49:54 +0000]
Bad test failed to show bug.
Tomas Doran [Fri, 17 Dec 2010 00:44:53 +0000]
Fix the case for body '0'
Tomas Doran [Thu, 16 Dec 2010 22:51:52 +0000]
Make response body able to be undef to allow RenderView to see 'defined but empty' body, allowing X-Sendfile to work nicer. Also removes horrible modifier code. win/win if it doesn't break anything else.
Tomas Doran [Thu, 16 Dec 2010 19:14:51 +0000]
Extra links in docs
Ronald J Kimball [Wed, 15 Dec 2010 16:44:50 +0000]
Add test case for uri_for() with #fragment and query params,
broken by 5.7008 when uri_for() was reimplemented without URI.pm.
Ferruccio Zamuner [Fri, 10 Dec 2010 10:22:41 +0000]
Found a another fault in chained action dispatcher.
Florian Ragwitz [Mon, 6 Dec 2010 12:12:45 +0000]
Merge branch 'master' into psgi
master:
Improved test about chained actions and add note in Changes.
Florian Ragwitz [Mon, 6 Dec 2010 12:08:49 +0000]
Don't build the full PSGI app more than once
Ferruccio Zamuner [Sun, 5 Dec 2010 23:08:01 +0000]
Improved test about chained actions and add note in Changes.
Tomas Doran [Sun, 5 Dec 2010 23:01:29 +0000]
TWMC issues were fixed, we need some more tests however
Tomas Doran [Sun, 5 Dec 2010 18:13:48 +0000]
Fix local_request. TODO - tests for local_request, make ->psgi_app lazy class data again
Florian Ragwitz [Sun, 5 Dec 2010 17:40:45 +0000]
Update TODO
Florian Ragwitz [Sun, 5 Dec 2010 17:35:08 +0000]
After startup, tell where the server is listening
Florian Ragwitz [Sun, 5 Dec 2010 17:29:13 +0000]
Set up MyApp->psgi_app lazily
This fixes the double inclusion through .psgi -> setup_psgi_app -> .psgi when
running with plackup.
Florian Ragwitz [Sun, 5 Dec 2010 15:05:16 +0000]
Separate raw psgi app and wrapped psgi app
Florian Ragwitz [Sun, 5 Dec 2010 13:40:15 +0000]
Fix merging mistake made in r13761
Engine::FastCGI is dead.
Florian Ragwitz [Sun, 5 Dec 2010 13:00:18 +0000]
Merge remote branch 'svn/trunk' into psgi
svn/trunk:
Rename --title to --proc_title as that makes more sense. Remove -t option as it's confusing..
Update changelog
Conflicts:
lib/Catalyst/Engine/FastCGI.pm
lib/Catalyst/Script/FastCGI.pm
t/aggregate/unit_core_script_fastcgi.t
Tomas Doran [Sun, 5 Dec 2010 12:33:40 +0000]
Rename --title to --proc_title as that makes more sense. Remove -t option as it's confusing..
Tomas Doran [Sun, 5 Dec 2010 12:22:56 +0000]
Update changelog
Florian Ragwitz [Sat, 4 Dec 2010 21:40:14 +0000]
Merge branch 'master' into psgi
master:
Stop passing along @INC
Use the proper way of setting @INC for tests
Aggregate http-server.t tests
Allow aggregating specific tests only
Use the harness instead of all of prove
Make the no-tabs test happy again
Florian Ragwitz [Sat, 4 Dec 2010 21:19:46 +0000]
Stop passing along @INC
It's pointless - it's always the same anyway.
Florian Ragwitz [Sat, 4 Dec 2010 21:19:41 +0000]
Use the proper way of setting @INC for tests
Florian Ragwitz [Sat, 4 Dec 2010 21:19:37 +0000]
Aggregate http-server.t tests
With this, I get to run the full suite in under 2 minutes again, as opposed to
the >3 minutes it took before.
Florian Ragwitz [Fri, 3 Dec 2010 17:46:45 +0000]
Allow aggregating specific tests only
Florian Ragwitz [Fri, 3 Dec 2010 17:46:40 +0000]
Use the harness instead of all of prove
Florian Ragwitz [Fri, 3 Dec 2010 16:00:19 +0000]
Don't build the psgi app twice
Florian Ragwitz [Fri, 3 Dec 2010 12:30:23 +0000]
Make the no-tabs test happy again
Florian Ragwitz [Fri, 3 Dec 2010 12:25:25 +0000]
Clarify what we really must do in order to merge
Florian Ragwitz [Fri, 3 Dec 2010 12:20:35 +0000]
Merge branch 'master' into psgi
master: (44 commits)
Fixed chained action order resolution to pass t/aggregate/live__component_controller_action_chained2.t and without break any other test.
Clarify end action after forward / detach
Fix passing file GLOBs to $c->res->body
Clarify docs
Added a --title parameter to fastcgi script, sets process title for FCGI::ProcManager
update MooseX module version due to deprecated warning
typo in docs fixed
Version 5.80029
Patch from the mailing list to clarify view warning
Avoid issues when just using CMOP::Class
Need to use _add_meta_method from new Moose onwards to avoid warnings
Actually bump versions, ACTUAL 5.80028, sorry
Version 5.80028
Bug fix for ctx_request, via matso++
Don't rely on qw() providing PAREN tokens
We use Class::MOP in Catalyst::Utils, so load it
Version 5.80027
Fix bug in the tests
Version 5.80026
Add doc patch for go and captures.
...
Conflicts:
lib/Catalyst/Engine/FastCGI.pm
lib/Catalyst/Script/FastCGI.pm
lib/Catalyst/ScriptRole.pm
lib/Catalyst/Test.pm
t/aggregate/live_component_controller_action_streaming.t
Ferruccio Zamuner [Sat, 27 Nov 2010 21:15:28 +0000]
Fixed chained action order resolution to pass t/aggregate/live__component_controller_action_chained2.t and without break any other test.
Tomas Doran [Tue, 23 Nov 2010 17:05:36 +0000]
Clarify end action after forward / detach
Stefan Seifert [Fri, 19 Nov 2010 12:08:17 +0000]
Fix passing file GLOBs to $c->res->body
Contrary to the documentation of Catalyst::Response, passing a
filehandle GLOB did not work, since the content-length header was not
computed correctly.
body checked if it was passed a blessed object and only then use stat
to get the file size. In all other cases the passed thing was considered
a string and length was used. Filehandle GLOBs are neither blessed nor
strings, so the content-length header always got set to 15 (length of a
stringified GLOB: "GLOB(0x123456)").
Tomas Doran [Mon, 15 Nov 2010 18:46:05 +0000]
Clarify docs