[
https://jira.codehaus.org/browse/WAGON-322?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=279663#comment-279663
]
Tamás Cservenák commented on WAGON-322:
---------------------------------------
Eh, you both are too fast! :D
So, yes, Benjamin has right, initially I was thinking about Reason-Phrase. But
after doing some research, I realized that current _Servlet API_ does not allow
you to customize easily the Reason-Phrase... furthermore, considering
frameworks like Jersey or Restlet, it goes almost next to impossible to set
those (I claim this without actually doing research in Jersey, it stands for
Restlet for sure).
So, the "extra header" sounds just as fine to me as the initially meant
Reason-Phrase.
So, I am open for discussion here....
> The HTTP Wagons should use Reason-Phrase sent in HTTP Response and use that
> in exception message (at least incorporate it)
> --------------------------------------------------------------------------------------------------------------------------
>
> Key: WAGON-322
> URL: https://jira.codehaus.org/browse/WAGON-322
> Project: Maven Wagon
> Issue Type: Improvement
> Reporter: Tamás Cservenák
> Assignee: Olivier Lamy
> Fix For: 2.0
>
>
> The HTTP Wagons should use Reason-Phrase sent in HTTP Response and use that
> in exception message (at least incorporate it).
> This would enable MRMs to at least "hint" somewhat Maven users why it's
> responding with some error.
> Typical example: Maven users using Nexus (but I believe is true for other
> MRMs too), tend to "forget" about having pull and push URLs different (you
> usually pull from "groups", and push/deploy to actual reposes). With wrong
> URL, your build will fail, and Maven message on CLI will state "server
> responded with code 400" (in case of Nexus, it will respond with Bad Request,
> since groups are read-only).
> Usually, the end user (and Maven) have no clue why the request was rejected,
> and user has to start investigating about the problem. On the contrary, the
> server side MRM actually knows very well what the problem is: deploy tried to
> read-only target, redeployment tried but redeploying is forbidden, deploy of
> a snapshot tried to a release repository, etc. All of these results with same
> error code (in case of Nexus): with HTTP 400, but without any explanation
> shown on client side.
> If Wagon -- I am not sure about the details -- would somehow incorporate the
> _actual_ Reason-Phrase got from server/MRM, and make Maven display it as part
> of the message about failed build, it would be much-much easier to realize
> why the build failed, it will be told by the MRM who actually rejected the
> request.
> By HTTP/1.1 RFC, the Reason-Phrase is unbound, it might have _any_ content
> (except CR LF), exactly for these cases.
> http://tools.ietf.org/html/rfc2616#section-6
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
