Re: String freeze break in GCC

[Christian Rose <menthos gnome org>]

mån 2004-02-16 klockan 18.29 skrev Kjartan Maraas:
> > Neither of those messages absolutely requires translation as they
> > are refering to the names of the browser appliction.   If people
> > prefer I'll reword them that way.  All of this code is somewhat in
> > flux anyway as that hard coded list is being moved into gconf.
> 
> I think that they should be marked and there's a good chance people will
> want to translate them anyway (at least the latter of the two). The
> challenge here is that every time people add strings translators have to
> update from CVS, update the po file, open editor of choice and look at
> the strings before deciding if it's worth translating. So the above
> argument doesn't necessarily mean there's no work involved on the
> translators side. Furthermore it affects stats for all languages :-)

Exactly. A freeze breakage is a freeze breakage. There's little
difference, at least for translators, between changes in an error
message that in the code will be displayed only once every fourth full
moon and a change in a message that will be plastered all over the
desktop with decimeter high bold block letters.

At least there's no technical or administrative way for translators to
differentiate between the two cases, so all freeze breakages are
unhelpful to the process of translation, especially once they become so
common that it becomes difficult to keep the track of the "important"
and the "unimportant" ones in all discussions about the string change
requests or freeze breakages du jour. Thus the need to limit *all*
string changes during freezes, no matter what code they're in.


> That said, I think this is a worthwhile change and we're still just
> entering the freeze period so I think we should fix as many of the
> remaining string-involving bugs as possible before entering hard freeze.
> 
> ...which leaves the question of having a slushy period before the freeze
> in the actual schedule so we can plan ahead when prioritizing these
> kinds of changes?

I think that if we go down the road of semi-official "soft" and "hard"
string freeze, we're in muddy waters. The freezes have been announced
many months in advance, so it should came as no surprise to anyone that
they're finally here. I'd like to still think of the current approvals
as exceptions, and would very much like to see the number of those
reduce drastically in the *very* near future.

I'm not saying that I like nor dislike the idea of official "soft" and
"hard" string freezes, just that testing that should not be for this
time around.


Christian