W dniu 19.10.2012 02:03, Michael Hudson-Doyle pisze:
I did not want to diverge the topic of the Connect discussion which may or may not be actually aiming at a construction of a unaffiliated test repository. The discussion in the wiki is specific to LAVA and I'd like to document what may be needed by various teams that I'm familiar with as well as explicitly open for others (in the weak term of comments which are somewhat less-than-perfect)
I think I agree with this, tbh. It's an interesting idea that may meet our needs, but we should concentrate on explicating our needs first.
I agree, hence more discussion on the problem than code/patches offering domain-specific solution.
Still, I'd like to keep this separate but open for discussion. Do you think you would consider working on a non-lava-specific test repository during the next cycle? If so then I would gladly move the spec to the test case management wiki that mwhudson created.
I think it depends on the nature of the solution. If the solution mostly consists of a protocol definition, I would think that the lava test repository would probably be a part of the same lava server process/database we already have. But not completely sure.
As for using or not using lava-server for this: I'm almost sure you'd need a part in lava-server to solve your goals but it does not exclude a third party system that can be shared by others. I think that a successful, pypi-like (single archive) test definition repository cannot depend on the existing lava infrastructure but I'm open to counter-arguments.
Thanks ZK