[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

(forw) Re: Launching a NMU campaign for pending l10n bugs

Thomas answer...

----- Forwarded message from Thomas Huriaux <thomas.huriaux@gmail.com> -----

Date: Fri, 15 Sep 2006 20:14:32 +0200
From: Thomas Huriaux <thomas.huriaux@gmail.com>
To: Christian Perrier <bubulle@debian.org>
Cc: Lucas Wall <lwall@debian.org>
Subject: Re: Launching a NMU campaign for pending l10n bugs
X-CRM114-Status: Good  ( pR: 30.2461 )


Christian Perrier <bubulle@debian.org> (15/09/2006):
> Quoting Lucas Wall (lwall@debian.org):
> > On 15/09/06 02:00, Christian Perrier wrote:
> > 
> > > I presented the procedure we used back in early 2005 and everybody
> > > agreed that we can re-use most parts of the process and
> > > infrastructure.
> > > 
> > > Do you think that we could wake up things and re-use that
> > > infrastructure? If you can't, we will of course setup something else,
> > > probably from your scripts....but of course, re-using it would be easier.
> > > 
> > > PS, later: see also Thomas Huriaux recent announcement in -i18n
> > 
> > Sure, no problem. The scripts would need some adjustments. Is the
> > current list the direct output of Thomas' script? Or does it have some
> > manual editing? Can the script be used to keep the packages status
> > updated? Beyond the package list (and some info focus changes) do you
> > think there are other changes needed?
> > 
> > If we iron out the details I'll work on the scripts over the weekend.
> Actually, let's try to get Thomas answer about all this and be sure
> that we're not doing duplicate work.
> Thomas, I suggest we work with Lucas so that he can update his old
> scripts to fit our new goal.....and you can adjust your extraciton
> scripts so that they keep the list used by Lucas script up-to-date as
> long as the work advances.

The current output of my script is only HTML (and does not need manual
editing). However, it would be really easy for me to create a database
in the format you think is appropriate, for example:

Package: foo
Score: 1234
Debconf: 333333, 333444, 444444
L10n: 222222, 222333

Package: bar
Score: 789

And similar entries for every package. Just tell me what would be your
best format and which information I should put in it. You could also
just take my script and copy/paste the interesting part to integrate
this ranking in your pages, so that there are no duplicate sites where
to find the information (or we could do it together).

The main difference that I see with the switch-to-po-debconf campaign is
that it should be done on a "package" basis and not on "bug" basis.
That is because of the important number of false-{positives,negatives}
due to the not-standardized title bugs used for po-debconf translations.
If a DD wants to NMU a package, he should pick one at the top of the
ranking, then check _every_ bug filled against the package and sort out
those which are po-debconf related and those which are not. If the DD
bases himself on the "po-debconf bugs:" section of the pages, he/she
will probably miss false-negatives and bugs where the l10n tag is

> It would also be nice if we get some progress report at the end of the
> campaign...thus following how many packages we've fixed and the
> "score" of each.

Just make a backup of the to-be-generated database at day-0 :-)

> I suggest we move this to debian-i18n so that discussions are
> archived. If you do so, please keep all quotes so that people can have
> a reference to originally private discussions. I don't do it now as I
> need you, Lucas, to be OK for initially private mails to be quoted
> publicly.

OK for me.


Thomas Huriaux

----- End forwarded message -----


Reply to: