Re: Q3 GNOME quarterly report

[Kenneth Nielsen <k nielsen81 gmail com>]

Den 09-11-2011 20:16, Gil Forcada skrev:
> El dc 09 de 11 de 2011 a les 16:47 +0100, en/na Kenneth Nielsen va
> escriure:
> > Den 09-11-2011 16:14, Nguyen Thai Ngoc Duy skrev:
> > > On Wed, Nov 9, 2011 at 6:50 PM, Kenneth Nielsen<k nielsen81 gmail com>   wrote:
> > > > Den 09-11-2011 09:45, Nguyen Thai Ngoc Duy skrev:
> > > > > Translation teams (or at least Vietnamese) tend to commit in bulk, the
> > > >
> > > > commit in bulk? across modules???
> > >
> > > I mean they are not small and beautiful commits like source code
> > > patches (or I'm too accustomed to git programming style in this
> > > regard).
> >
> > Ahh, then I understand. But as long as we agree on that, commits are
> > still a somewhat useful activity metric. If you are updating a fully
> > translated desktop and ends up with a fully translated desktop, then a
> > commit should roughly equal the following number of strings:
> >
> > number_of_updated_strings_per_cycle/(number_of_modules *
> > number_of_updates_per_module_in_a_cycle)
> >
> > For non-100% languages there should also be a growth factor in there
> > somewhere.
> >
> > But the point is, that as long as non of those numbers changes
> > dramatically from one cycle to the next, number of commits should still
> > make a reasonable number for translator activity _comparison_ between
> > cycles.
> >
> >
> > In any case, if we want a real number bad enough. I should certainly be
> > possible to pull the po-file from before and after the commit and use a
> > podiff tool to count the changes.
>
> Actually my idea was mostly to get a general feeling on how much do we
> commit in general and maybe on per language, that way if we start seeing
> less and less commits from the French team we can ask them about that or
> trying to help them in some sense.
>
> And also, as the quarterly reports are meant to be an overview of each
> team, this kind of metrics are quite useful in general terms (obviously
> not as a qualitative/quantitative metric for the translation project
> team).

I think we are talking a bit past each other. My point with the first 
point was exactly, that the number of commits should make a perfectly 
reasonably metric for _comparisons_ of activity, which can be used for 
the kind of "general terms" measurements that you are talking about.

The last part was just general ramblings about how precise metrics could 
be achieved, in case that is required for this purpose or in general.

Regards Kenneth