[petsc-dev] Open PRs to be merged for release 3.7

Barry Smith bsmith at mcs.anl.gov
Sun Apr 24 13:00:32 CDT 2016

> On Apr 24, 2016, at 12:28 PM, Jed Brown <jed at jedbrown.org> wrote:
> Satish Balay <balay at mcs.anl.gov> writes:
>> I was hoping for a feature freeze - by my maint branch merge.
>> [and Barry agreed to that :). See the text of the post: "[petsc-dev] plans for PETSc release"]
> I think the text of that email inadvertently (or intentionally?) changed
> the release process in a rather fundamental way that I disagree with.
> We never had a petsc-dev saying that a feature freeze will be in effect
> starting Xday and dumping 'master' into 'maint' without a feature freeze
> is reckless because we're knowingly advertising more stability (via the
> 'maint' label) than is actually present.

    The only mistake was Satish could have called the "freeze" branch, for example maint-3.7 or something  instead of maint, but that IMHO is not very important (as Satish noted any die hard fan of maint who wants to stick to the previous release can just switch to maint-3.6 which they would have to do anyway as soon as the release is made, so why is that a big deal.)
>> Irrespective the maint issue - the changes are not yet tested in
>> next. Even if its tested in tonights next - they would be untested in
>> maint/master.
>> So I can't  do both - merge them tomorrow and spin a tarball tomorrow.
> My preference would be to push the release date back by a few days so we
> can ensure a more stable v3.7.  This irrespective of whether Lisandro's
> changes are merged for v3.7.

   Unfortunately for political reasons we have to push out a release now regardless. Of course fixes will go into a patch release soon.

   Since Lisandro's changes are backward compatible (and his code has been very clean recently) I think we should just take a chance and put it in maint. This isn't mission critical software that will collapse civilization if there is a problem somewhere.


More information about the petsc-dev mailing list