[
https://issues.apache.org/jira/browse/MNG-6795?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17284993#comment-17284993
]
Daniel Compton edited comment on MNG-6795 at 2/16/21, 1:31 AM:
---------------------------------------------------------------
I'm also a maintainer of clojars.org and separately run
[Deps|https://www.deps.co], a private Maven repository service. I use the
reason phrase to help users debug errors more clearly with Deps.
I fully agree that using the HTTP reason phrase to pass this information was a
gigantic hack. If it was being proposed today as a new feature, I would vote
against it given that it has been removed in HTTP/2 and was never intended for
this use. However, it's a vital feature used by many public and private Maven
repository managers to help users and customers debug problems.
bq. Another problem is that Wagon is protocol agnostic, this means that even if
we add those information to Wagon exceptions, it remains unused for other Wagon
implementations except HTTP.
I can see that this is not ideal, but it's also the current situation that
we're in. I would guess that the majority of Maven users are using HTTP Wagons,
and the overwhelming majority of errors occurring using Maven would happen with
an HTTP Wagon.
I don't have any terribly strong feelings about the exact implementation
details, I'm happy to update my code to use whatever is the new format.
was (Author: danielcompton):
I'm also a maintainer of clojars.org and separately run
[Deps|https://www.deps.co], a private Maven repository service. I use the
reason phrase to help users debug errors more clearly with Deps.
I fully agree that using the HTTP reason phrase to pass this information was a
gigantic hack. If it was being proposed today as a new feature, I would vote
against it given that it has been removed in HTTP/2 and was never intended for
this use. However, it's a vital feature used by many public and private Maven
repository managers to help users and customers debug problems.
> Another problem is that Wagon is protocol agnostic, this means that even if
> we add those information to Wagon exceptions, it remains unused for other
> Wagon implementations except HTTP.
I can see that this is not ideal, but it's also the current situation that
we're in. I would guess that the majority of Maven users are using HTTP Wagons,
and the overwhelming majority of errors occurring using Maven would happen with
an HTTP Wagon.
I don't have any terribly strong feelings about the exact implementation
details, I'm happy to update my code to use whatever is the new format.
> define a replacement for ReasonPhrase to display details about transfert
> failures
> ---------------------------------------------------------------------------------
>
> Key: MNG-6795
> URL: https://issues.apache.org/jira/browse/MNG-6795
> Project: Maven
> Issue Type: Task
> Components: Artifacts and Repositories
> Affects Versions: 3.3.9, 3.6.3
> Reporter: Herve Boutemy
> Priority: Major
> Fix For: 4.0.x-candidate
>
>
> as documented in WAGON-541
> [372aa52c80c2f7b458c095d837e15ebeafbdf0b7|https://gitbox.apache.org/repos/asf?p=maven-wagon.git&a=commit&h=372aa52c80c2f7b458c095d837e15ebeafbdf0b7],
> ReasonPhrase has been removed from HTTP/2 and deprecated (optional) in
> HTTP/1.1 for more than five years. It is useful to give detailed error
> messages on artifact transfert errors with a repository manager.
> A replacement has to be found
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
