[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-i18n-doc
Subject:    Re: Having the merge/scatter script on a kde server for French locale
From:       Chusslove Illich <caslav.ilic () gmx ! net>
Date:       2019-04-21 17:56:57
Message-ID: 201904211956.58259.caslav.ilic () gmx ! net
[Download RAW message or body]


> [: Albert Astals Cid :]
> * I see summit as a "non official" thing mainly because it is
> maintained manually (i think) (by Chusslove (i think))

In the meantime, it seems to me that Luigi too knows how to handle day-
to-day stuff. Here I mean for the gather operation, other operations are
done by language teams. So it's currently a bus factor of say... 1.8?

As for making summit gather run automatically on a KDE server (which
Luigi too mentioned recently and in the past), a major problem there is
that often a manual intervention is needed (for mapping catalogs and
processing orphans). This happens 1-3 days per week.

> * AFAIK Summit is kind of CPU intensive, which is understandable you'd
> like to run it automatically in KDE's servers, but then it should be
> for everyone (that wants to run summit) and i'm not sure at this point
> we have enough CPU for that.

Summit gather is done once per day on templates, with run times:

  (circa 2010 machine) Athlon II 3.0 GHz: 24 min
  (circa 2018 machine) Ryzen 3.2 GHz: 15 min

Only one core is used.

Summit merge and scatter is done by each language team with arbitrary
periodicity. Run time of these operations depends heavily on what
special things the team has configured to be done. E.g. if someone set
fuzzy matching on merge from a compendium, it could take hours. Maybe
someone who is running a mostly default merge/scatter could report on
this? E.g. Italian and Korean don't look too menacing, maybe you could
report some run times?

-- 
Chusslove Illich (Часлав Илић)

["signature.asc" (application/pgp-signature)]

[prev in list] [next in list] [prev in thread] [next in thread]