Florian Ragwitz [Wed, 13 Jan 2010 07:59:18 +0000]
Depend on a ReverseProxy middleware that spits out secure envs for requests forwarded to secure ports.
Tomas Doran [Tue, 12 Jan 2010 01:30:26 +0000]
Do the right thing with FCGI args, the hostname and the keepalive options in the scripts so that the Plack servers get them correctly.
There are still a load of options for the dev server (like writing a PID file) that we'll need to add to the plack server (or do ourselves manually?), and the entire set of options we actually pass into the application via ->run is unused by it.
I'm not sure if we are safe to change the original option passing convention or not - need to look at alternate engines and stuff..
Tomas Doran [Mon, 11 Jan 2010 23:36:52 +0000]
Move everything back out into the scriptrole, allow the scripts to force their engine..
Florian Ragwitz [Mon, 11 Jan 2010 22:23:56 +0000]
Canonicalize the url becuse we're building it from the always-undecoded REQUEST_URI, not PATH_INFO.
Florian Ragwitz [Mon, 11 Jan 2010 22:23:49 +0000]
Well, apparently this *is* necessary, just the extra_env stuff isn't :/
This reverts commit
87d4377bb66a6530cd2e13c4e770e7b1f0aaf7d3.
Florian Ragwitz [Mon, 11 Jan 2010 22:12:16 +0000]
Delete unfixable FAIL.
It's unnecessary under PSGI anyway. The PSGI servers won't let get undecoded
paths through to us.
Florian Ragwitz [Mon, 11 Jan 2010 21:41:38 +0000]
Port t/aggregate/live_engine_request_escaped_path.t to psgi.
Florian Ragwitz [Sun, 10 Jan 2010 23:22:18 +0000]
Skip Engine::CGI-specific tests until we find a better place for them.
Florian Ragwitz [Sun, 10 Jan 2010 20:30:06 +0000]
At least pass the port param to _run_psgi_app to get http-server.t running again.
Florian Ragwitz [Sun, 10 Jan 2010 19:05:43 +0000]
Stop expecting the Host header for proxy requests to be the original one.
Doing this was probably stupid in the first place, and now we're handing all
the frontend proxy stuff off to plack, we shouldn't even care anymore. The
middleware is free to set HTTP_HOST to HTTP_X_FORWARDED_HOST for us.
Florian Ragwitz [Sun, 10 Jan 2010 17:03:50 +0000]
Start supporting running behind reverse proxies again.
Florian Ragwitz [Sun, 10 Jan 2010 17:03:32 +0000]
Inline the OverrideEnv middleware into Catalyst::Test.
Florian Ragwitz [Sun, 10 Jan 2010 06:31:16 +0000]
Improve handling errors during prepare.
Florian Ragwitz [Sun, 10 Jan 2010 06:31:03 +0000]
Look up the method just once.
Florian Ragwitz [Sun, 10 Jan 2010 04:42:21 +0000]
Properly merge the extra env var into the psgi environment.
Florian Ragwitz [Sun, 10 Jan 2010 04:42:14 +0000]
Don't dump psgi.input as it can't be deserialized.
Florian Ragwitz [Sun, 10 Jan 2010 04:22:01 +0000]
Adapt to the changed _customize_request calling convention.
Florian Ragwitz [Sun, 10 Jan 2010 04:09:59 +0000]
Properly set REMOTE_USER in the request env.
Florian Ragwitz [Sun, 10 Jan 2010 04:09:52 +0000]
Add a way to specify additional psgi values in the psgi env via Catalyst::Test.
Tomas Doran [Sun, 10 Jan 2010 00:33:41 +0000]
Use Plack::Loader and push the running of plack back into the engine code. Unsure if here or scriptrole is the best place for it, thoughts? (48 => 13 fails though)
Tomas Doran [Sun, 10 Jan 2010 00:00:43 +0000]
Delete code. 64 => 48 subtests failed
Florian Ragwitz [Sat, 9 Jan 2010 21:21:21 +0000]
Remove outdated comment and some trailing whitespace.
Florian Ragwitz [Sat, 9 Jan 2010 21:16:46 +0000]
Fix typo.
Tatsuhiko Miyagawa [Sat, 9 Jan 2010 21:16:40 +0000]
depend on Plack 0.903 for streaming interface support in Plack::Test
Tomas Doran [Sat, 9 Jan 2010 19:37:56 +0000]
Require Plack..
Florian Ragwitz [Sat, 9 Jan 2010 18:12:51 +0000]
Clear up the response callback and the writer object after each request.
Florian Ragwitz [Sat, 9 Jan 2010 18:01:25 +0000]
Don't pass $ctx down to read().
Florian Ragwitz [Sat, 9 Jan 2010 17:49:24 +0000]
No, don't warn in the log. clients being retarded don't concern us usually.
Florian Ragwitz [Sat, 9 Jan 2010 17:49:17 +0000]
Let read_chunk read from psgi.input.
Florian Ragwitz [Sat, 9 Jan 2010 17:49:11 +0000]
There's no Content-Length for streaming responses.
Florian Ragwitz [Sat, 9 Jan 2010 17:49:05 +0000]
Remove other engines.
Florian Ragwitz [Sat, 9 Jan 2010 17:48:57 +0000]
Clear the request env after finishing a request.
Florian Ragwitz [Sat, 9 Jan 2010 17:48:51 +0000]
Delay erroring out on wrong usage of Catalyst::Test.
We should probably deprecate
use Catalyst::Test;
(without an app name) in favour of
use Catalyst::Test ();
Florian Ragwitz [Sat, 9 Jan 2010 17:48:45 +0000]
Tell the writer object when we're done.
Florian Ragwitz [Sat, 9 Jan 2010 17:48:38 +0000]
Start porting Catalyst::Test to Plack::Test.
Florian Ragwitz [Sat, 9 Jan 2010 17:48:31 +0000]
Work towards supporting psgi.
Florian Ragwitz [Tue, 5 Jan 2010 00:28:35 +0000]
Stop supressing Adopt::NEXT warnings.
Tomas Doran [Mon, 4 Jan 2010 20:55:25 +0000]
Clarify comment
Tomas Doran [Mon, 4 Jan 2010 20:52:10 +0000]
Doc fix
Tomas Doran [Wed, 30 Dec 2009 15:55:43 +0000]
Fix regex special characters screwing things up by not using regexes
Tomas Doran [Tue, 22 Dec 2009 14:19:36 +0000]
Someone think of a less fugly way of doing this please? Fixes using rewrite rules to ask for a sub-path in your app with apache in some combinations..
Tomas Doran [Fri, 18 Dec 2009 19:55:03 +0000]
Fix RT#52898, __MOP__ removal breaking debug screen with C::P::Session
Tomas Doran [Tue, 15 Dec 2009 10:21:32 +0000]
Bug fix for issue reported on the mailing list by Toby Corkindale
Florian Ragwitz [Fri, 11 Dec 2009 23:28:09 +0000]
moar better warnings fix.
Tomas Doran [Fri, 11 Dec 2009 23:26:08 +0000]
Changelog, bump version
Tomas Doran [Fri, 11 Dec 2009 23:25:35 +0000]
Remove warning when running tests aggregated
Tomas Doran [Fri, 11 Dec 2009 12:44:55 +0000]
Unfuck that as well
Tomas Doran [Thu, 10 Dec 2009 09:46:10 +0000]
Fix RT#52630
Tomas Doran [Wed, 9 Dec 2009 21:36:13 +0000]
Additional test
Tomas Doran [Wed, 9 Dec 2009 20:54:53 +0000]
More tests for the prepare_path thing, fix said tests, changelog
Tomas Doran [Wed, 9 Dec 2009 20:50:18 +0000]
Changelog
Tomas Doran [Wed, 9 Dec 2009 20:35:13 +0000]
And more tests and fixing for the same thing
Tomas Doran [Wed, 9 Dec 2009 20:21:13 +0000]
Fixes port environment, RT#52604
Tomas Doran [Wed, 9 Dec 2009 18:30:53 +0000]
Regression for 5.80015 when rewriting an app, nanonyme++, Khisanth++
Tomas Doran [Tue, 8 Dec 2009 11:16:44 +0000]
Fix tests from r12252, add comments to make the behavior explicit
Dave Rolsky [Tue, 8 Dec 2009 05:39:25 +0000]
Doc --host change
Dave Rolsky [Tue, 8 Dec 2009 05:38:40 +0000]
Don't default to localhost for --host
Tweak docs for --host to say it accepts a name or IP
KMX [Sun, 6 Dec 2009 20:11:43 +0000]
Win32 fix: reverting commit 12210 (after a discussion with t0m) - all tests pass on Win32/strawberry perl 5.8.9 and 5.10.1
Dave Rolsky [Sun, 6 Dec 2009 18:50:57 +0000]
Doc change for --restartdirectory
Dave Rolsky [Sun, 6 Dec 2009 18:48:34 +0000]
fix pod for Server script so options match actual attr name
Dave Rolsky [Sun, 6 Dec 2009 18:44:36 +0000]
Add restartdirectory as alias for restart_directory, for backcompat
Florian Ragwitz [Sun, 6 Dec 2009 16:31:52 +0000]
Undocument $request->user.
Tomas Doran [Sun, 6 Dec 2009 13:04:50 +0000]
Skip on win32, hopefully we'll get a better answer than this, but lets stop failing
Tomas Doran [Sun, 6 Dec 2009 12:47:59 +0000]
Win32 fix, kmx++
Tomas Doran [Thu, 3 Dec 2009 17:15:17 +0000]
I can't stand the over-long debug screens any more. Suggestions on how to do this in a less gross way?
Gerda Shank [Wed, 2 Dec 2009 21:52:21 +0000]
add another failing test for Chained CaptureArgs preference
Tomas Doran [Wed, 2 Dec 2009 19:13:19 +0000]
And that's what I get for not really applying the patch and doing it manually. Fail..
Tomas Doran [Wed, 2 Dec 2009 18:58:14 +0000]
Pod nits, rt#52370
Tomas Doran [Wed, 2 Dec 2009 15:10:52 +0000]
Changes, bump version
Tomas Doran [Tue, 1 Dec 2009 04:19:44 +0000]
The documentation example had the variables the opposite way round to my working nginx config. CLEARLY this meant the documentation needed the variables (but not their values) transposing. Yes, yes - I really am _that_ dumb.
Tomas Doran [Tue, 1 Dec 2009 02:16:08 +0000]
Fix bug in Catalyst::Engine which could cause it to all go wrong if read returned '0' as per thread on mailing list re 'Wrong Content-Length value' and clarify docs about read and read_chunk methods in Catalyst::Engine
Tomas Doran [Tue, 1 Dec 2009 00:54:23 +0000]
Bump deps and changelog for another dev release with new scripts and scary path resolving changes so I can have another call to test.
Tomas Doran [Tue, 1 Dec 2009 00:52:57 +0000]
TODO out gshanks chained tests
Tomas Doran [Mon, 30 Nov 2009 23:23:38 +0000]
Rewrite changelog entry to be more complete
Tomas Doran [Mon, 30 Nov 2009 23:20:12 +0000]
Merge 'trunk' into 'fix_path_info_decoding'
r12121@t0mlaptop (orig r12086): t0m | 2009-11-30 22:59:15 +0000
Require new M::I
Tomas Doran [Mon, 30 Nov 2009 23:17:33 +0000]
Require new HTTP::Request::AsCGI
Tomas Doran [Mon, 30 Nov 2009 22:59:15 +0000]
Require new M::I
Tomas Doran [Mon, 30 Nov 2009 22:05:41 +0000]
Back out r12083, also clarify the docs - like this works for me now, and the bottom para seems the wrong way round...?
Tomas Doran [Mon, 30 Nov 2009 21:02:14 +0000]
I think the nginx hackery can just die with the new code in ::Engine::CGI
Tomas Doran [Mon, 30 Nov 2009 20:44:16 +0000]
And yea, these test cases now pass
Tomas Doran [Mon, 30 Nov 2009 18:26:34 +0000]
Commit miyagawa's fix for PATH_INFO decoding. I think that this also implies changes / fixes in the fcgi _fix_env stuff
Tomas Doran [Sun, 29 Nov 2009 17:16:16 +0000]
Merge 'trunk' into 'better_scripts'
r13204@nc10 (orig r12073): rafl | 2009-11-29 16:43:17 +0000
Depend on a recent leakchecker to stop everything from crapping over itself if PadWalker is too old.
Florian Ragwitz [Sun, 29 Nov 2009 16:43:17 +0000]
Depend on a recent leakchecker to stop everything from crapping over itself if PadWalker is too old.
Tomas Doran [Sun, 29 Nov 2009 15:39:41 +0000]
Merge 'trunk' into 'better_scripts'
r13191@nc10 (orig r12060): gshank | 2009-11-27 22:43:11 +0000
testcase for Chained bug - dispatches to wrong action
Florian Ragwitz [Sun, 29 Nov 2009 15:33:55 +0000]
Tidy things up a little and fix pod usage.
Florian Ragwitz [Sun, 29 Nov 2009 15:33:48 +0000]
This is not a release.
Florian Ragwitz [Sun, 29 Nov 2009 14:00:51 +0000]
Less trailing whitespace.
Florian Ragwitz [Sun, 29 Nov 2009 14:00:43 +0000]
Remove useless lexical.
Florian Ragwitz [Sun, 29 Nov 2009 14:00:38 +0000]
Port numbers need to be > 0.
Florian Ragwitz [Sun, 29 Nov 2009 14:00:31 +0000]
Tidy up some attr definitions.
Florian Ragwitz [Sun, 29 Nov 2009 14:00:25 +0000]
Properly validate restart regexen.
Florian Ragwitz [Sun, 29 Nov 2009 14:00:19 +0000]
Make follow_symlinks more consistent with the other restarter args.
Florian Ragwitz [Sun, 29 Nov 2009 14:00:12 +0000]
Depend on a new CMOP because we rely on its new load_class error messages.
Tomas Doran [Sat, 28 Nov 2009 12:38:12 +0000]
Adding the files you need for the tests you've committed helps them pass, idiot :)
Gerda Shank [Fri, 27 Nov 2009 22:43:11 +0000]
testcase for Chained bug - dispatches to wrong action
Tomas Doran [Fri, 27 Nov 2009 02:08:33 +0000]
Bump version in runtime branch
Tomas Doran [Thu, 26 Nov 2009 23:40:27 +0000]
Pass app name into constructor
Tomas Doran [Thu, 26 Nov 2009 22:46:26 +0000]
Merge 'trunk' into 'better_scripts'
r12057@t0mlaptop (orig r12022): t0m | 2009-11-26 01:21:22 +0000
Back out 11979 until I work out why IPC::Run is getting hold of the FCGI file handle
r12070@t0mlaptop (orig r12035): t0m | 2009-11-26 21:29:32 +0000
Fix test on bleadperl RT#52100
Tomas Doran [Thu, 26 Nov 2009 21:29:32 +0000]
Fix test on bleadperl RT#52100
Tomas Doran [Thu, 26 Nov 2009 21:02:57 +0000]
Fix typo, bricas++, add simple test for restarter arguments