I agree. I created PRs to have build running on Jenkins (and be sure it’s green 
before merging).

If you want to be reviewer on some, please let me know.

Thanks
Regards
JB

> Le 9 mars 2020 à 14:15, Christopher Shannon <[email protected]> 
> a écrit :
> 
> Furthermore the regressions in STOMP can be seen here in Jenkins:
> https://builds.apache.org/view/A/view/ActiveMQ/job/ActiveMQ-Java8/lastCompletedBuild/testReport/
> 
> Also looking at another fix that was just merged a couple days ago:  I'm
> not convinced that this is correct:
> https://issues.apache.org/jira/projects/AMQ/issues/AMQ-7314  There's no
> test here to demonstrate the fix and it seems like it might be wrong to
> mark a sequenceId less than the lastDeliveredSequenceId to be the
> lastDeliveredRef.  A change like this needs to demonstrate the issue and
> fix works
> 
> My recommendation when doing future releases is to freeze the release
> earlier in the process and to not try and merge in a ton of stuff in the
> last couple days before a release.  Last minute changes can be pushed to
> the next release unless critical/blockers.  Merging at the last second
> before a release doesn't give much time for others to review the changes
> and verify they make sense or to even give CI a chance to run and be
> verified as shown here.
> 
> On Mon, Mar 9, 2020 at 8:49 AM Christopher Shannon <
> [email protected]> wrote:
> 
>> -1, there are multiple STOMP regressions here
>> 
>> 1) In some of my tests that verify error handling works properly, I'm
>> getting a NPE here that causes my build to fail:
>> https://github.com/apache/activemq/blob/activemq-5.15.12/activemq-stomp/src/main/java/org/apache/activemq/transport/stomp/ProtocolConverter.java#L261
>> 
>> The root cause is there's no guarantee there will be a linked exception
>> and in this case the linked exception is null leading to an NPE
>> 
>> 2) It makes no sense to swallow the exception here and not propagate it
>> up:
>> https://github.com/apache/activemq/blob/3c302dce33500cd8e36f626af333ce208ebc44a0/activemq-stomp/src/main/java/org/apache/activemq/transport/stomp/StompNIOSSLTransport.java#L67
>> 
>> Every other protocol throws the IOException and this no longer does.
>> Furthermore, it makes no sense to treat STOMP different than other
>> protocols.  I don't see why we need to log at all when the IOException
>> should propagate up and then be logged elsewhere.  If it's not being logged
>> properly then something else up the chain (TcpTransport or whatever) should
>> probably be catching the exception and logging.
>> 
>> On Mon, Mar 9, 2020 at 8:33 AM Daniel Kulp <[email protected]> wrote:
>> 
>>> +1
>>> 
>>> Dan
>>> 
>>> 
>>>> On Mar 6, 2020, at 2:15 AM, Jean-Baptiste Onofre <[email protected]>
>>> wrote:
>>>> 
>>>> Hi everyone,
>>>> 
>>>> I'm submitting ActiveMQ 5.15.12 release to your vote.
>>>> 
>>>> This release includes dependency updates (especially for CVE/Security
>>>> fixes), important improvements on JDBC persistence adapter, other
>>> improvements and fixes.
>>>> 
>>>> Please take a look on the Release Notes for details:
>>>> 
>>>> 
>>> https://issues.apache.org/jira/secure/ReleaseNote.jspa?projectId=12311210&version=12346500
>>> <
>>> https://issues.apache.org/jira/secure/ReleaseNote.jspa?projectId=12311210&version=12346500
>>>> 
>>>> 
>>>> The Maven staging repository is:
>>>> 
>>> https://repository.apache.org/content/repositories/orgapacheactivemq-1204
>>>> 
>>>> The dist staging repository is:
>>>> https://dist.apache.org/repos/dist/dev/activemq/activemq/5.15.12/ <
>>> https://dist.apache.org/repos/dist/dev/activemq/activemq/5.15.12/>
>>>> 
>>>> Git tag:
>>>> activemq-5.15.12
>>>> 
>>>> Website PR:
>>>> https://github.com/apache/activemq-website/pull/28 <
>>> https://github.com/apache/activemq-website/pull/28>
>>>> 
>>>> Please vote to approve this release:
>>>> 
>>>> [ ] +1 Approve the release
>>>> [ ] -1 Don't approve the release (please provide specific comments)
>>>> 
>>>> This vote will be open for at least 72 hours.
>>>> 
>>>> Thanks !
>>>> Regards
>>>> JB
>>> 
>>> --
>>> Daniel Kulp
>>> [email protected] <mailto:[email protected]> - http://dankulp.com/blog <
>>> http://dankulp.com/blog>
>>> Talend Community Coder - http://talend.com <http://coders.talend.com/>
>>> 
>> 

Reply via email to