dolfin team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

ok, it would be good to also hear Garth's thoughts.

If there is no enthusiasm for a big release this week, I suggest that
we make a release of 0.9.12 tomorrow (or as soon as the new
variational problem interface is working). Any other remaining bugs or
blueprints can be retargeted to rc-1 which we can work on in August.
That way we can generate and archive the documentation for 0.9.12 on
the web page, which will be a good test for the documentation/web.

What does "deployment" mean that you have listed for one of your
blueprints?

--
Anders


On Tue, Jun 28, 2011 at 01:26:51AM -0700, Johan Hake wrote:
> Not sure what is best, but I have at least finished bugs and blueprints
> assigned to me for both 0.9.12 and 1.0.0-rc1. I will also not be available for
> code sprint this week.
>
> Johan
>
> On Monday June 27 2011 03:38:11 Anders Logg wrote:
> > B0;268;0cDear all,
> >
> > What are your thoughts on a release of 1.0? One of the main obstacles,
> > at least for me personally has been the FEniCS book which has now been
> > submitted.
> >
> > Is there any interest in a code sprint this week, to try to have
> > something ready for the release by the end of the week? I'm up for it.
> >
> > I see two different options:
> >
> > 1. Merge the milestons 0.9.12 and 1.0.0-rc1 and release 1.0.0-rc1 at
> > the end of the week. Then we collect (and maybe fix) bug reports
> > during the summer and aim for a release of 1.0 in August (possibly
> > after a 1.0.0-rc2 and rc3).
> >
> > 2. Finish up and release 0.9.12 this week and then go into release
> > mode in August with 1.0.0-rc1, 1.0.0-rc2, ..., 1.0.0.
> >
> > In both cases (after releasing 1.0-rc1) we should only fix bugs (not
> > add new features or change the interface) before releasing 1.0.0.
> >
> > I don't know when the Debian import freeze is, so it may have
> > implications on the choice we need to make.
> >
> > Another thing to discuss is what should happen after 1.0.0. I think it
> > would be good to be much more conservative with interface changes than
> > what we have been. With the latest change to VariationalProblem, I
> > think we have converged pretty well so I don't foresee any big changes
> > will be needed.
> >
> > This also relates to the policy in Debian for binary compatibility
> > with shared libraries which may prevent any big changes to the
> > interface. I think Johannes knows more about this.
> >
> >
> > So (1) or (2)? Or none of the above? In either case, I think we need
> > to make a common decision so we can coordinate and others know what to
> > expect.
> >