[ooo-build] Do you consider Go-oo project to be a fork of OpenOffice.org?

[Kohei Yoshida]

On Thu, 2010-10-07 at 00:54 +0100, espinosa_cz wrote:
> > I would really be interested in knowing how this analysis was done
> (though the 
> > png says 'active developers' rather than commits).
> 
> There is an extensive article to it:
> http://people.gnome.org/~michael/blog/ooo-commit-stats-2008.html
> 
> By the way, I cannot see much of Michael Meeks recently.
> He used to be more active here, he was quite "face" of go-oo to me, 
> his blogs was often cited as was critical of SUN handling the project.
> I expect him to announce Go-oo to become (finally) a fork, to became 
> LibreOffice.

He already did:
http://people.gnome.org/~michael/blog/2010-09-28.html

> > Anyways, newer analysis would surely be preferred. 
> 
> Absolutely. I was hoping that someone from the active developers here 
> could point us to something newer.

Well, I can tell you that a new analysis will be unlikely, and I'll tell
you why.

The original analysis was done by analyzing the commit log of the
then-used CVS repository.  We wrote a script to parse the CVS commit
logs to pick up relevant commit information to generate the data which
was later analyzed and presented in Michael's blog.  I originally wrote
that script myself, and it took about 3 or 4 days to bring that script
to a usable state.  Also, each script run would take about 30 minutes to
parse the entire CVS history and generate data.

Then the OOo project migrated to Subversion, and a year later to
Mercurial.  So, to do the same analysis all over again, a new script
would need to be written to handle the new repository type.  Also, they
threw away quite a bit of commit history in favor of saving disk space,
so the new analysis would not be as accurate as we want it to be.

> Such analysis is important now, if Go-oo/LibreOffice developers count 
> for less then say 20% of OOo new code then I am quite concerned 
> about the project future.

You need to keep in mind that there are lots of code that have not been
integrated into the OOo repository.  So, analyzing the OOo repository
alone would give you a wrong picture of reality.  Plus, a simple code
count alone is not conclusive of any project's future.  It's one metric
to measure project's well being, but not the only metric that matters.

BTW, I encourage you to join the libreoffice at lists.freedesktop.org to
discuss stuff like this.  This list is deprecated, and hardly anyone
pays attention to this list.

Kohei

-- 
Kohei Yoshida, LibreOffice hacker, Calc
<kyoshida at novell.com>