On 01/15/2013 01:23 PM, Skip Montanaro wrote:
I know about the technical advantages of moving from centralized to
distributed version control. That, by itself, won't change anything.
Correct me if I'm wrong, but I don't think that the suggestion of moving
to git*HUB* has anything to do with centralized vs distributed version
control. That wouldn't make any sense because SF has git too.
I think it has everything to do with the fact that SF is intended for
*project* hosting and GH is for *code* hosting.
If I may give some links:
http://usersinhell.com/why-sourceforge-lost/
http://readwrite.com/2011/06/02/github-has-passed-sourceforge (from 2011!)
http://williamedwardscoder.tumblr.com/post/24944468513/how-github-slew-sourceforge-and-the-rest
(talks about forking is bad vs forking is good)
Besides, Sourceforge supports Git, so we could, in theory, convert
from svn to git and leave the code at SF. Does GitHub offer some sort
of "help wanted" or publicity features that make it likely that those
pull requests will just start happening?
I would suggest keeping the project site on SF (to keep using the
"publicity features") and moving the code base to GH (to take advantage
of social coding). Heterogeneity is the new norm. Use the best tool for
the job.
or in the words of SF itself: "Many people use us together; GitHub for
collaboration, and SF.net for distribution"
(http://sourceforge.net/blog/github-collaboration-and-haters/)
Wait a second... no I don't suggest doing that. I suggest that the
person who first proposed it, takes the initiative. Something something
about talking and walking...
In fact, code migration to GH is so easy that I have already done it
myself: https://github.com/amedee/spambayes
The only thing that I need, is to fill in the correct email addresses in
an authors-transform.txt:
anadelonbrin = anadelonbrin <anadelonb...@spambayes.org>
anthonybaxter = anthonybaxter <anthonybax...@spambayes.org>
bkc = bkc <b...@spambayes.org>
bwarsaw = bwarsaw <bwar...@spambayes.org>
gvanrossum = Guido van Rossum <gu...@python.org>
gward = gward <gw...@spambayes.org>
hooft = hooft <ho...@spambayes.org>
htrd = htrd <h...@spambayes.org>
jhylton = jhylton <jhyl...@spambayes.org>
jvr = jvr <j...@spambayes.org>
kpitt = kpitt <kp...@spambayes.org>
mhammond = mhammond <mhamm...@spambayes.org>
montanaro = montanaro <montan...@spambayes.org>
nascheme = nascheme <nasch...@spambayes.org>
(no author) = spambayes-dev <spambayes-dev@python.org>
npickett = npickett <npick...@spambayes.org>
popiel = popiel <pop...@spambayes.org>
richiehindle = richiehindle <richiehin...@spambayes.org>
rubiconx = rubiconx <rubic...@spambayes.org>
sjoerd = sjoerd <sjo...@spambayes.org>
tim_one = tim_one <tim_...@spambayes.org>
timstone4 = timstone4 <timsto...@spambayes.org>
uid26747 = uid26747 <uid26...@spambayes.org>
xenogeist = xenogeist <xenoge...@spambayes.org>
(Everybody knows Guido, right? That's the only address that I am 200%
sure about.)
For now I just have added @spambayes.org after each username but don't
worry, this is trivial to fix afterwards with git filter-branch.
If the 23 other developers should also have a GH account then we might
see some interesting charts like this:
https://github.com/amedee/spambayes/graphs/code-frequency
As you can see, most of the work was done before 2005. Then there was a
big spike of activity early 2007, but since then, nothing much has happened.
Another observation: all work was done in one branch. I'm not used to
that any more. :)
--
Amedee
_______________________________________________
spambayes-dev mailing list
spambayes-dev@python.org
http://mail.python.org/mailman/listinfo/spambayes-dev
