[petsc-dev] PETSc developers who use weird MPI

[Jed Brown]

Satish Balay <balay at mcs.anl.gov> writes:

> On Fri, 1 Nov 2013, Barry Smith wrote:
>>    I made the mistake of believing “it is ok to have some branch hang around unmerged with master etc for several months” myth
>> 
>
> This mode works better when there are no dependencies between multiple
> branches.

When you merge someone else's branch into your branch, you get their
features as well as their bugs.  So if they fix those bugs and you need
the bugs to be fixed, you need to merge their branch again.  All else
being equal, we want to minimize the amount of "unstable" stuff from
other branches, so that we don't have to deal with their bugs.

Integration through 'next' increases the chances that their bugs will be
found and fixed within that branch, before their branch is merged to
'master' (at which point the branch with all its features and bugs will
be part of all new development).

Merging a branch to 'next', or offering it for someone else to merge
into their branch, should be thought of as a making a micro-release,
with both the assertion of expected stability and duty to fix bugs in
the release.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 835 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20131101/79daabed/attachment.sig>