BDFL decision of Python's DVCS

[Matthew D. Fuller]

On Tue, Mar 31, 2009 at 05:27:32PM +0200 I heard the voice of
Matthieu Moy, and lo! it spake thus:
> Marius Gedminas <marius at pov.lt> writes:
> 
> > If you care about one more person's impression, it's that "Bazaar
> > is expected to change dramatically for the better in the next
> > couple of months" has been true (or at least claimed to be true)
> > for the last year at least.  After some point I stopped believing
> > the promises.
>
> This is IMHO the kind of promises that harmed bzr a lot.

Unfortunately, that sort of thing has a long traditional presence,
both with claims about the future exceeding what actually appears, and
claims about the present overstating the case   :|

It goes back at least as far as when we were on the cusp of the weaves
-> knit switch.  Of course, the saddish thing there may be that the
change could well have been the oft-claimed "30 times performance
improvement".  My daily pulls of the weave bzr.dev could take upwards
of 30 minutes.

And when pack-0.92 came in, and we were going to introduce a series of
4 new pack formats based on that.  Here's the mail from Oct 31.
Adjusting the indexes, replacing the inventory, changing the
compression format, various other smaller changes.  With one new
format per release after 0.92.  It looks like brisbane-core will
finally actually contain pretty much all of the substantive changes
there that were proposed to be done by 0.96.


> "We expect 1.0 to release by March 2007 with performance equivalent
> to the best in the field."

And there WERE performance numbers that showed bzr 1.0 equal with or
besting the best in the field, on wall time performance and disk space
usage.  For add/commit of single-rev imports of large projects with no
history.



-- 
Matthew Fuller     (MF4839)   |  fullermd at over-yonder.net
Systems/Network Administrator |  http://www.over-yonder.net/~fullermd/
           On the Internet, nobody can hear you scream.