On Tue, Dec 05, 2006 at 03:02:32PM +0100, Anders Logg wrote:
On Tue, Dec 05, 2006 at 01:33:29PM +0100, Johan Jansson wrote:
I've created a branch of DOLFIN called DOLFIN-stable. Since every
repository (every time you clone for example) is also a branch, this
is nothing dramatic. The only difference is that this branch is
publically available in the same way as DOLFIN is.
The purpose of this branch is to keep the kernel stable. The only
kernel changes allowed are backports of bugfixes and possibly
important overlooked functionality from the DOLFIN
repository.
However, module changes are allowed as before. Kernel development is
supposed to happen in the DOLFIN repository (where applications need
to be stable to test the kernel) and application development is
supposed to happen in the DOLFIN-stable repository (where the kernel
needs to be stable to test the application).
At regular intervals (at least at DOLFIN releases), the DOLFIN and
DOLFIN-stable branches will be merged. Thus the DOLFIN and
DOLFIN-stable branches will be equivalent at the time of every
release.
Who does the merge?
The maintainer of dolfin or the maintainer(s) of dolfin-dev?
And who maintains what? Jansson maintains dolfin-stable. Garth/me
maintains dolfin?
I will be the primary maintainer of dolfin-stable
(applications/modules), and I guess the logical thing is for you and
Garth to be primary maintainers of dolfin (kernel). I will still be
involved in the kernel however, and I expect everyone else to still be
involved in the modules as well.
I think a reasonable labor division is that I have the final
responsibility for the merge, and you (Anders and Garth) have the
final responsibility for release? This does not of course forbid
helping out, but there needs to be someone who steps in when things do
not happen by themselves.
How does that sound?
