[oi-dev] Git as a version control system for new OI projects
buffer.g.overflow at googlemail.com
Wed Jun 22 20:43:31 UTC 2011
-----BEGIN PGP SIGNED MESSAGE-----
I may be making a huge leap here, but I had some expectation that the tools issue are navigable if Joyent's been using git (I'll immediately concede that there are all kinds of ways I might be wrong about that). I don't mean to diminish the question of toolchain transition issues across SCMs, but I think there's a thought process that's not just "ooh, shiny" underlying the suggestion to move to git, which should be foregrounded.
As you rightly say, there's a viability threshold that has to be passed or cleared, but if that's sorted, I think the questions shift from what needs to be supported by integration with the existing toolchain to understanding how branching and such are supposed to work as processes, what of that isn't already integrated into the toolchain (i.e. because that's direct interaction with the SCM), and whether there's a substantial improvement in implementing that workflow with a particular SCM. The problems we've got with things like MQ are encountered precisely where the toolchain ends and people discover that operations they have to perform in the SCM are a bit of a hole in the head. It could be that we're underutilising Mercurial, whether in its core feature set or its extensions, and it would definitely help evaluate that if the use cases and our understanding of the issues were spelled out.
I don't think we get to the right answer without having both workflow/process and existing integration points on the table, but we may never get to any question about toolset integration if we are clear about the former.
On 22 Jun 2011, at 18:50, Garrett D'Amore wrote:
> My biggest concerns about switching to git really relate to tooling. If
> the tools can all accommodate git, then I don't really care.
> - Garrett
-----BEGIN PGP SIGNATURE-----
Version: GnuPG/MacGPG2 v2.0.12 (Darwin)
-----END PGP SIGNATURE-----
More information about the oi-dev