openerp-community-leaders team mailing list archive
-
openerp-community-leaders team
-
Mailing list archive
-
Message #00050
Re: Simple things we need from Tiny for better bug planning/management
On Friday 22 January 2010, you wrote:
> Although I personally also prefer git over bazaar I think we need to take
> into account several things:
It currently is only the practical side (I don't claim it to be the ultimate
solution), but I may have some answer to your concerns:
> - Though a good version control system can improve productivity I don't
> think git alone solves the problem we have at the moment.
If we lower the "cost" of maintaining the code, our resources could be
exploited better, into real development.
One interesting point is that merges (the ones that are non-trivial and
contain conflicts) can be /outsourced/ . This means that, even if Tiny are the
sole mainteners of the "stable" branch, they can always fetch an appropriately
done merge from any partner that dared do that.
If you add some testing tools (oor, the "base_module_check" perhaps?), then we
automate much of the release procedure and reduce the resources needed for our
regular work.
> - Bazaar has slightly better support on Windows than git has (at least that
> was like this some time ago, please correct me if I'm wrong). Not that it
> can't be used, but it's easier to use bazaar than git in that platform. I
> don't use Windows very much but there are people that do..
Support for Windows-based development is a waste of resources, IMHO, because
we impose a procedural overhead to *all* developers, just because windows are
crippled and some people can't abandon them. I'm not paying the windows "tax",
for one.
> - Moving everything to git at the moment would be very painful, the problem
> is there are no resources and making such a change would require lots of
> resources.
There is already some solutions to that: I have long used the 'tailor' tool,
which is trivial. The newer 'git-fast-import' one should fully cover us. For
one, I can help and undertake this task.
All this thread is still *not* a suggestion that we immediately move over. But
everybody is free to consider this solution as a remedy to our current
situation. I believe my public branches could be a starting point for anybody
that wants to test this technology, already.
References