[
https://issues.apache.org/jira/browse/CXF-6515?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14680663#comment-14680663
]
Sergey Beryozkin edited comment on CXF-6515 at 8/10/15 7:55 PM:
----------------------------------------------------------------
Hi Romain - the bus related update goes in the next day or two.
Let me clarify once again. I argued a lot about: 1) if the custom provider
matches then do not even check the default ones 2) more recently - if a default
provider had a non-wildcard @Produces then it does not mean another provider
with a wildcard is less specific because it can check a media type in
isWriteable/isReadable, but well, in the end of the day, I had to do this
change, I think it is positive in its own way that CXF will now pass some
remaining TCK tests it used to fail otherwise, and the workaround about some
situations are possible.
We have a custom comparator - I'm still curious why you see some strange
results there :-), please investigate. The bus update will go shortly. The
custom providers which extend AbstractConfiurableProvider can do a very
specific Consumes/Produces and the runtime will respect that. Narrowing down
Produces/Consumes is most likely a needed compromise...Overall I reckon we can
manage the situation well...
was (Author: sergey_beryozkin):
Hi Romain - the bus related update goes in the next day or two.
Let me clarify once again. I argued a lot about: 1) if the custom provider
matches then do not even check the default ones 2) more recently - if a default
provider had a non-wildcard @Produces then it does not mean another provider
with a wildcard is less specific because it can check a media type in
isWriteable/isReadable, but well, in the end of the day, I had to this change,
I think it is positive in its own way that CXF will now pass some remaining TCK
tests it used to fail otherwise, and the workaround about some situations are
possible.
We have a custom provider - I'm still curious why you see some strange results
there :-), please investigate. The bus update will go shortly. The custom
providers which extend AbstractConfiurableProvider can do a very specific
Consumes/Produces and the runtime will respect that. Narrowing down
Produces/Consumes is most likely a needed compromise...Overall I reckon we can
manage the situation well...
> provider sorting not right
> --------------------------
>
> Key: CXF-6515
> URL: https://issues.apache.org/jira/browse/CXF-6515
> Project: CXF
> Issue Type: Bug
> Reporter: Romain Manni-Bucau
> Assignee: Sergey Beryozkin
> Fix For: 3.1.3, 3.0.7
>
>
> Didnt dig too much into it but seems sorting with a custom comparator is not
> respected, can be linked to 6514, if this 6514 is fixed I'll test again to
> confirm this one
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
