Re: Moving to bzr?

["Stephen J. Turnbull" <stephen@xemacs.org>]

John Yates writes:

 > That defaulting to rich-root may take bzr some time does that preclude
 > the emacs project from initializing its master repository as rich-root?
 > After that would not rich-rootiness simply propagate to new branches?

It's not rich-root itself that is most important, though rich-root is
indeed an important feature for the Emacs repository.  It's a
heuristic for "work on the repo format/other important aspects has
proceeded sufficiently far."  There are several changes in process
that promise to allow dramatically improved performance on very
frequent operations like "bzr log".  Some of those may need a new repo
format.

I will note that Python is currently creating a PEP to move the Python
repositories from Subversion to a DVCS.  The three candidates are bzr,
hg, and git.  Despite a strong prior bias toward bzr, the PEP
proponent's initial impression upon actually trying some of the
scenarios is a shocked "bzr is almost unusable, might kill off one-off
contributions from new contributors" [my paraphrase].  I expect that
will be moderated with experience in more scenarios, but if somebody
with a strong prior that bzr is a good way to go reacts in that
fashion, I think we have to worry that potential Emacs contributors
will do so, too.

Please do read the initial impressions section at the end of
http://docs.google.com/View?docid=dg7fctr4_40dvjkdg64.  I'm
deliberately overstating the case ... maybe. :-(  The GNOME DVCS
survey also showed bzr as quite weak in comparison to the other two
leading candidates:
http://blogs.gnome.org/newren/2009/01/03/gnome-dvcs-survey-results/

I do *not* intend to reopen a debate on which DVCS to use; that is
decided.  I do intend a caution about timing.  In terms of the tools
that people use daily, "fools rush in where angels fear to tread" and
"better the devil you know than the devil you don't" are proverbs we
should not take lightly.