Produce first declared content-type on Accept:*/*

[mads-b]

This is a fix to reported issue JERSEY-2635, stemming from
the incorrect ordering of for loops in MethodSelectingRouter.

[jerseyrobot]

Can one of the admins verify this patch?

[mpotociar]

Hi, thank you for the contribution.

In order to accept your contribution, we need you to sign Oracle Contributor Agreement, see here for details: https://jersey.java.net/scm.html#/Submitting_Patches_and_Contribute_Code

[mpotociar]

Jenkins, please test this patch.

[mpotociar]

Jenkins, please test this patch.

[mpotociar]

Hi, we're still waiting for your signed OCA. Have you sent it to us yet?

[mpotociar]

Seems like there is a problem with a copyright header in one of the changed test classes:
https://jersey-ci.ci.cloudbees.com/job/github-pull-request-verification/95/artifact/copyright.year.wrong.txt

[AdamLindenthal]

Mads-b, could you please change the copyright header year in the ContentNegotiationTest to 2013-2014 instead of 2013 and re-commit?

Thanks!

[AdamLindenthal]

Mads-b, are you still here? Are you planning to sign the OCA and fix the header or should we rather consider to commit internally?

Regards,
Adam

[mads-b]

I am truly sorry for my extended silence. Tldr: my company is hesitant to sign such an agreement. If possible, I would appreciate it if this patch was merged internally. If it makes a difference, I do not need to retain any rights at all to this patch, I merely want it to be present in future Jersey releases.

[AdamLindenthal]

Hi, thanks for letting us know. We can either close the pull request now, leave the JIRA issue still valid and look at the issue in one of the next sprints internally, or you can still consider signing the agreement yourself as an individual (this makes much more sense if you would plan to make further contributions in the future).

Just let us know once you have decided.
Thanks,
Adam

[mads-b]

You can't copy this patch, make the necessary changes and merge it internally? I fixed this because we really needed this behaviour. I doubt I'll contribute again if no more severe bugs are found on our part.

[shamoh]

Hi Mads.
I think we can take your change without your commit history.
I would like to ask what was the reason you changed methodBuilders type from Set to List? Was there some special reason?
Thanks,
Libor

[mads-b]

I'm not entirely sure, since it was a while ago since I wrote this, but I recall fixing two bugs in succession: The ordering of the values in @produces annotation, but also an issue that surfaced after fixing this. For some reason, having multiple resource methods producing different media types results in random methods getting invoked when using the Accept: / header, when the topmost method should be invoked. The reason it works without the patch is pure coincidence, but after fix, the problem became more prominent.

[shamoh]

I've been 'inspired' by your code. Fix will be in 2.16 release, I hope. Thanks again for your contribution.