On 10/14/13 2:06 AM, "Erik de Bruin" <e...@ixsoftware.nl> wrote:
>> I think (even though it's a little painful) we have go with the tests >>passing as tagged (even though they are not part of the source release) >>otherwise a new RC is needed. Otherwise there's no way of reproducing >>the test results accurately. In this case carrying over the previous >>votes would apply. > >In case of a new RC where I could test everything in a fresh copy, >without modification, I'd most likely change my vote to +1. For the record, I do not test against the branches or tags. I expand the source package and build it, then set FLEX_HOME to point to it, then from my git repo I run mustella. IMO, that's a better test than building and testing from the repo and tags because the source-package contains a subset of files from the repo and we want to make sure there is nothing missing. The output from mustella tells me it is picking up the flex-config from the expanded source package so I know I'm getting those SWCs. Now there could be a flaw in that logic, but that's how I've been testing. And that's why I can vote +1. There is nothing required to change in the RC1 source package. We could have someone temporarily revert Maurice's DG fix and build an RC2 and it would be identical to RC1. And if we do decide to do an RC2 to pick up these optional changes then I would spot test a few things and carryover my vote. I would also diff the RC2 against RC2 to verify that only the things I spot tested are different, but that's a lot less work than tying up two machines for 10 hours to run full mustella on them. -Alex
