[
http://jira.amdatu.org/jira/browse/AMDATU-283?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=11123#comment-11123
]
Bram de Kruijff commented on AMDATU-283:
----------------------------------------
Ivo> - In this refactoring mimetype resolving is handled by Felix instead of
Sling. I doubt if this is the right choic...
Not sure if we have a use case for the sling mimetype support. Made an
improvement issue (AMDATU-371) to deal wit this question outside critical path
Ivo> WAR deployment seems to be broken for context paths other then ''. ...
Contextpath handling and requestwrappers where poorly implemented. Resolve
resulted in latest major refactor under r950
Ivo> There seems to be no unit test nor integration test that actually uses
MimeTypeResolver or SecurityHandler, so this is untested.
Added a few. Some more for dispatcher/httpcontext would be appropriate. Will do
that later
Ivo> init/destroy in AbstractHandler is not implemented properly w.r.t.
concurrency; init and destroy may be invoked simultaneously.
This class is part of a composition. Concurrency is guarded at the service
callback level.
Ivo> more concurrency issues in ResourceProviderListener; reentrant read/write
locks have been added in the past to resolve...
Saw the ad-hoc check and removed it. If we can reproduce we should file a bug
on dependencymanager
Ivo> if you stop the http context bundle, Amdatu halts/stops completely
You mean hangs right? Removed a deadlock in the servlethandlerregistry
Ivo> stop/start the dispatcher bundle causes many errors and stacktraces
Shindig/Wink stuff not service lifecycle aware. Added dependencies.
Ivo> Add javadoc to public interfaces (i.e. to ResourceProvider,
MimeTypeResolver, SecurityHandler)
Added some. Needs mechanism description on WIKI
Ivo> Copyright statement is missing on some files
Added some.
Ivo> Add a WIKI page
Ivo> Update the WIKI
Will do docs later under general wiki cleanup
Open consideration;
* Merge httpcontext into dispatcher? They don't really do anything usefull
without eachother and seem
rather coupled in terms of releasecycle. Now developer typically need to depend
on both artifacts.
* Publish JSP/Resource support alias keys on central interface? Now developer
typically need to depend
on both artifacts.
* ServletHandler target matching now is eager. FilterHandler matching should
probably be to according to
spec and would allow additional caching.
* Needs much more testing, requires refactor to get interfaces out.
> Refactor web httpcontext mechanism
> ----------------------------------
>
> Key: AMDATU-283
> URL: http://jira.amdatu.org/jira/browse/AMDATU-283
> Project: Amdatu
> Issue Type: Improvement
> Components: Amdatu Web
> Affects Versions: 0.1.0
> Reporter: Bram de Kruijff
> Assignee: Bram de Kruijff
> Fix For: 0.2.0
>
> Attachments: AMDATU-283-v1.patch
>
>
> As discussed and documented at
> http://lists.amdatu.org/pipermail/amdatu-developers/2011-January/002474.html
> the current HttpContext requires some refactoring to be really usefull and
> support advanced use cases (including security).
> Ad. Discussed HttpContext with multiple backing bundles on OSGi-Dev just to
> be sure. No showstoppers :)
> http://www.mail-archive.com/[email protected]/msg01930.html
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
_______________________________________________
Amdatu-developers mailing list
[email protected]
http://lists.amdatu.org/mailman/listinfo/amdatu-developers
