0.8.3 bugfix release

[Wouter van Heyst]

On Sun, Jun 18, 2006 at 02:00:56PM -0500, John Arbash Meinel wrote:
> Wouter van Heyst wrote:
> > Quite a lot of people join #bzr wondering what broke when they push to a
> > new location and bzr reports that is has pushed zero revisions. When
> > they check their public http location, they see an empty dir. Explaining
> > this has become a bit of a routine, but perhaps it is worth to put out a
> > new bugfix release containing the fix for this? r1714 of bzr.dev merged
> > robertc at robertcollins.net-20060515050657-7232e2d1f6895fae , which should
> > do the trick.
> > 
> > Wouter van Heyst
> 
> Well, there are quite a few bugfixes that have been committed that would
> be worthy of another release.
> I think that we could argue that as long as we don't change any data
> formats, then we could release it as another point release.
> So we could release something like the pycurl + multirange request
> change, which speeds up a 'get' by ~2.5x.
> The only thing I'm hesitant to release as a point release is my encoding
> work, because it has test cases which don't pass on Mac. Its because it
> is testing functionality that has been broken, but it would be nice to
> have all official releases pass all tests.
> Though it has been a long time since win32 was there. I've done cleanups
> from time to time (as has Aaron and Alexander). But it still needs quite
> a few more rounds of testing and cleanups.
> 
> I would love to get the tests passing on Win32 and OS X, and then
> release an 0.8.3 with a bunch of our speed improvements and general
> bugfixes.

Due to a Debian point of view I am used to point releases with minimal
changes, but I wouldn't mind a compatible point release with speedups
and no regressions.

Wouter van Heyst