Re: Different string totals in status

["Raphael Higino" <raphaelh uai com br>]

> >> When you notice such a thing, I suggest you report it to Bugzilla
> >> for the relevant component, and CC some/all of the i18n hackers
> >> around who might be able to help fix the issue (such as
> >> menthos menthos com, , chris gnome-de org,
> >> dsegan gmx net ...).
> >
> > I notice this every day and I would have to write a Bugzilla report for
> > every status update, or almost.
>
> Well, this happens because some modules don't even use intltool
> properly, and thus can't be avoided.
>
> Also, you'd need not "write a Bugzilla report for every status
> update", you'd more likely need to write a Bugzilla report for
> several applications for which you notice such a behaviour (and most
> of them already have bug reports in place).  Usually, this is not a
> status pages fault (sometimes it is, but generally not), so report
> bugs against applications instead.
>
> > I'd have to compare the amount of all languages strings. Then, if I
wanted
> > to know in what packages are the miscounted strings I'd have to get into
> > each language status and compare it to the others. It sound like a
script's
> > work. Maybe it should be done along the status update, and that's why
I'm
> > CC-ing Carlos (if I'm not wrong, he takes care of the status page
update).
>
> Well, if you're willing to go around fixing such stuff, perhaps
> Carlos can provide you with detailed error log of every run of status
> pages (he already sends that to Abel I believe, so it shouldn't be a
> problem).
>
> OTOH, if you're not willing to do this, I suggest you either report
> single instances you *do* notice (like the Evolution case) to Bugzilla,
> and let others handle whatever they come across.  That's how we have
> worked so far, and it turned out to be acceptable.

Sorry for coming back to this subject and bothering you after more than a
month, but today when I was reading the thread "Pig latin stats have
appeared" I remembered it.

That's the thing. I went to the link
http://l10n-status.gnome.org/gnome-2.8/pig-latin/extras/ in that thread just
to see how it looked that "language" status page. I found out something
interesting: some modules' graphs in the extras and fifth-toe's status page
for the pig-latin "language" appear in yellow, so they don't have strings to
be translated. They are:

fifth-toe: balsa, gcompris, gtetrinet, totem.
extras: eggcups, epiphany-extensions, gdesklets-extras, gftp, giftoxic,
gimp-tips, gnome-commander, gnome-mud, gnome-themes-extras, jumpnbumpmenu,
libmimedir, sodipodi.

I tried to understand why those packages appeared in yellow here, when they
should appear in red, and (here is the interesting and weird thing) when I
clicked in sodipodi to open its POT file there was no POT file for it, as
you can se in http://l10n-status.gnome.org/gnome-2.8/PO/, and so there are
not for the others, although they have strings to be translated and there
are teams which already translated them.

I think this may be the reason for why there is that difference in the
totals (around 5000 strings when we take the languages with most and least
strings).

Why does this happen? Is this a problem with the module or with the updates?
What should I do about this? Should I report a bug for each of this modules?

I think this thing with POT files is a big problem (or almost). There are no
translations for eggcups, for instance, to Brazilian Portuguese because the
POT file just doesn't show and is not on the status page (and I just can't
access the GNOME-CVS from where I translate so I could generate them). I
just discovered that eggcups has translatable strings today.

What you think?

Raphael Higino