Re: [VOTE] Release Apache qpid-interop-test 0.2.0 (RC2)

[Robbie Gemmell <ro...@gmail.com>]

-1. There are some blocking issues needing resolved.

I spent a chunk of time yesterday and again today trying to get the
tests going, I've raised some blocking/critical bugs and some
improvements noticed while doing that, and had some general comments
in addition.

The 'utils' module is incorrectly versioned as 0.1.0:
https://issues.apache.org/jira/browse/QPIDIT-130

The tests reference the 0.1.0 version of the Qpid JMS 'shim' and so
none of those tests work at all:
https://issues.apache.org/jira/browse/QPIDIT-129

Misusing JAVA_HOME makes the tests error without a clear indication
why when its set:
https://issues.apache.org/jira/browse/QPIDIT-128

I actually mentioned this one last year before 0.1.0 went out and
noted it again in discussion elsewhere not that long ago. Some of the
other issues (I tried running against Artemis) I ran into masked me
realising that I was still hitting this one. Its not technically a
blocker, but is effectively. Either way it should be fixed.

Some other issues or improvements that dont affect 0.2.0:
https://issues.apache.org/jira/browse/QPIDIT-131
https://issues.apache.org/jira/browse/QPIDIT-127

Some general comments:

We stopped distributing MD5 checksums several months ago, that needs deleted.

Having -rc2 in the filename means you will have to modify the files
and checksums before final release. Its nice if the dist-dev artifacts
dont need changed so that the final step is just a remote svn copy,
removing risk of messing up any changes (which has happened
previously).

Using a branch called 2.0.0 is going to make for annoyance later, when
there ends up being a tag and branch with the same name, I'd suggest
changing the branch to 2.0.x.

I had to follow the note about using "--exclude-shim ProtonPython3"
from line 315 of the QuickStart, because my source install of Proton
0.24.0 only had the Python 2 bits. Theres a bunch of detection at
install for whether particular things are present before enabling
their tests, should this (and vice-versa) maybe be one of them? It
seems weird for it to run those tests and wait (10?) seconds for them
to error when they could never work.

It seems off that the first thing in the README is note of the source
repo. Someone reading it most likely either downloaded the release and
dont need the repo, or are looking at it from the repo being
described. We also dont direct regular users to unreleased stuff as a
matter of course, so the repo being the first thing in the README
seems off.

As mentioned before 0.1.0, one nice to have thing (though it seems
simple enough to do): The JSON[-P] dependencies used are permittable
,but would need called out as CDDL if we ever made a bundle, so it
seems like it would be nicer to consider using e.g ALv2 bits from
Apache Johnzon and geronimo-specs.

Robbie

On 30 June 2018 at 17:24, Kim van der Riet <ki...@redhat.com> wrote:
> Hi,
>
> I have assembled a proposed release for Apache Qpid Interop Test 0.2.0
> (RC2). The issues found in RC1 were fixed. Please try it out and test it,
> then vote accordingly on this thread.
>
> The files can be obtained from:
> https://dist.apache.org/repos/dist/dev/qpid/interop-test/0.2.0-rc2/
>
> The JIRAs included in this release are:
> https://issues.apache.org/jira/secure/ReleaseNote.jspa?projectId=12318621&version=12342084
>
> This release is tagged in git as 0.2.0-rc2.
>
> Thanks,
> Kim
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@qpid.apache.org
> For additional commands, e-mail: users-help@qpid.apache.org
>

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@qpid.apache.org
For additional commands, e-mail: users-help@qpid.apache.org