You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@uima.apache.org by Marshall Schor <ms...@schor.com> on 2015/03/03 00:11:43 UTC

Re: [VOTE] [CANCEL] Release UIMA SDK 2.7.0 rc7

I thought about this, and feel it's better to just keep this as 2.7.0. 

I do agree that the release votes can't by themselves, do great QA - that has to
come from
our processes and JUnit tests, and early adopter testing, where possible.

-Marshall

On 2/27/2015 2:46 PM, Richard Eckart de Castilho wrote:
> I think ther have been some quite significant changes at the lowest levels that
> deserve extensive testing. The release votes are unable to provide this kind of
> quality assurance. 
>
> Maybe it might even be worth to do a 2.7.0-beta-1?
>
> Cheers,
>
> -- Richard
>
> On 27.02.2015, at 20:02, Marshall Schor <ms...@schor.com> wrote:
>
>> A user reported an issue - see https://issues.apache.org/jira/browse/UIMA-4267. 
>> This is an older problem - it occurs in 2.6.0 as well as 2.7.0.  It probably
>> hasn't been reported before because it only manifests itself for fairly large
>> use cases (more than 6 million items in the map).
>>
>> The fix is pretty small, but it's a blocker for this user's (and of course,
>> other potentially "large" users') use; so I'm reluctantly concluding that it's
>> worth rolling back the release, doing the fix, and doing rc8 (a record, I think
>> :-) ).
>>
>>
>> -Marshall
>
>