Hi,

> No objection here.  What do others think?  I thought it was a common
> practice to put tests in the same repo as the code

In my experience unit tests yes, integration tests not always.

> How will it help the vast majority of folks who just want to use the latest
> code?  Create a branch and get going but let’s also get this release out
> now.

Part of the voting +1 on a release is that a PMC member must be able to
compile it. [1] Currently that’s not the case for me, although I’ve not tried
the latest dev branch.

IMO we also are like to get quicker feedback on releases and find
issues earlier if we  have a build that's easier for people to compile
and play about with.

> Well, good luck with that.  I don’t think there is a policy or other
> mandate that builds have to be on ASF machines

The policy is that the release artefacts needs to be signed on machines
under the control of a single committer who has exclusive access to it
and explicitly not on another machine (ASF or otherwise). [2][3]

However it is preferred that ASF infrastructure is used for testing / CI as 
that 
is a supported platform. Infrastructure have said they will give us an Azure 
box.

Thanks,
Justin

1. http://www.apache.org/dev/release.html#approving-a-release
2. http://www.apache.org/dev/release-distribution.html#sigs-and-sums
3. http://www.apache.org/dev/release.html#owned-controlled-hardware



Reply via email to