Hi Thomas,
Thomas Lange <la...@cs.uni-koeln.de> wrote (Wed, 29 Nov 2023 20:13:50 +0100):
> Because these are important pages with many hits, I would expect to
> have these translation updated in a few weeks, maybe 1-2 weeks.
>
> When should we remove these outdated translations?
> What's your oppinion?
I think we cannot focus thus strict on these "main pages".
We have to lift the view and look on the whole field of work.
There are some very active translators and those languages are fine, but
the majority of languages suffer from a lack of manpower IMO (as many teams
in Debian do).
And we are all volunteers; volunteers in difficult times, with an
(apparently) endless sequence of crisis, wars, and all the problems they
implicate.
Even if one might think, that does not affect Debian directly, I think it
affects the amount of work those volunteers can dedicate to Debian work.
Or it can prevent them from Debian work at all, due to technical or
personal reasons.
So, the point is: translator's time might be rare; and you cannot force
them on what pages they work, there might be more effective criteria then
just the number of hits in our statistic.
Another point might be (while this is my personal opinion here !!!):
there are too much commits to English, which only change trivial things
(remove blank lines; fix typos in English; fix wording ...) and the
translations are not synced or cannot be synced by the author of the
English change. And the translations (might) become outdated because of this.
That leaves unnecessary work to translators, or at least they might think
so ("unnecessary") about such changings. And that affects their motivation
and usage of their time.
We probably need to be more careful with changings in English for trivial
reasons.
Now I remember the big overhaul of the website done by Heike Jurzik
(~2 years ago?).
Due to the lack of manpower in the www team, we hired a professional
web designer, doing the changings in full time, which resulted in heavy
changings on the website.
Did anyone think about hiring professional translators, to sync the
translations? I guess not. The amount of changings were way to high for
normal translator's time slots, so they most probably will need years to
catch up with those changings!!!
Or they give up and leave the pages alone...
And for usual changings:
I think we need to much resources (on the English side as well as on
translator's) for repeating tasks like pages under ../releases.
Basically, the files under
../releases/buster
../releases/bullseye
../releases/bookworm
../releases/trixie
are nearly the same, but with small differences in codenames, release
dates and such.
But nevertheless all those pages need to be touched by translators
many times over the release cycle (if translators want to keep their
files up-to-date).
Even pages for oldstable need translator's time to be up-to-date
(I just spotted exactly this situation for Bullseye today!)
And remember: if one changes a page in English, that might imply work for
dozens of translators!!!
I can think about changing the whole mechanism behind these pages, to create
a basis, which has content for all situations during the whole release
lifetime, without any need to change that underway:
So, we would create such "template" for English, translators work on
translating the relevant parts and then they have their template for
their language as well, and when a new release comes, we just run a
sed script, which changes the codenames from the stable one to oldstable,
testing to stable, and so on. The release dates need to be adjusted,
but with that most of the work should be done.
Another relevant situation is when LTS period starts or ends for a
release, but I imagine we can get this done be just changing an entity,
which is used for all languages, and we are done.
Most probably it's not that easy as I write it here, I'm aware of that.
It's just a rough idea at the moment.
But I would like to propose this, because IMO it's worse it to better use
translator's time slots.
Sorry Thomas, long story here, and I did not answer on your initial
question yet :-)
I would not vote too strict against the removal of outdated translations,
in fact I did that some years ago for German as well, due to lack of
manpower in the German team.
You don't get applause for such initiative, it's an unthankful job.
That being said, I could imagine a timeframe from ~6 months for this
to be senseful?
On the other hand, it occurred to me, if it's possible, to only change the
filename from let's say index.wml to index.wml.old instead of removing the
file (assuming that the wml build process of the website ignores such
files; did not check that).
That would make it very easy for translators, to catch up with their work,
if they find time.
Of course you might say "Hey, the file is not lost, we have a git repo
here! No need for such trick."
That's of course correct, but translators might not be as familiar with
such advanced usage of git as DDs are.
So I think it would be worse it.
What do you think?
(Should I stop writing here?
Yes, this mail has gone long enough for today :-) )
So long
Holger
--
Holger Wansing <hwans...@mailbox.org>
PGP-Fingerprint: 496A C6E8 1442 4B34 8508 3529 59F1 87CA 156E B076
