20.08.2018 15:43 Matthew Miller <mattdm(a)fedoraproject.org> wrote:
On Fri, Aug 17, 2018 at 10:06:54PM +0200, Rafal Luzynski wrote:
> This seems easy to measure if we are able to count the number of
> downloads of specific packages, here langpacks would be interesting,
With our current setup, we're not able to do that.
That's unfortunate but I think I understand the reasons (see below).
Thank you for your answer.
Some more replies to other people:
19.08.2018 19:00 "pravin.d.s(a)gmail.com" <pravin.d.s(a)gmail.com> wrote:
On the other side.. translations definitely helps to gain more contributors.
It is low barrier contribution. Once new contributor learn and become
contributor to Fedora he can explore other domains.
Definitely it is helpful but my question is how much it is helpful.
IIRC this question was asked in the context of Filipino language 
but personally I am also worried about languages like Lower Sorbian 
which I recently helped to include in Linux (I mean Linux in general,
not just Fedora). It has about 5000-7000 native speakers, most of them
are older people and all of them speak fluently German. How likely is
it that anybody will ever use Fedora in Lower Sorbian?
20.08.2018 16:53 Frau Silvia Sánchez <lailahfsf(a)gmail.com> wrote:
Oh, shame. I was getting all enthusiast about the idea. Any suggestions to
find out the same but maybe using a different method? Is it possible?
I'm afraid there is no different method. The measurement must be
performed on the Fedora repository servers, in this case on the
mirror servers. If I understand correctly, currently a Fedora machine
connects with the main repository server and sends its OS version
and architecture. This is the moment where these statistics can be
collected. Then the main repository server replies with the list of
mirror servers. The Fedora machine connects with a mirror server
and downloads required packages. The main server does not get any
information about what packages (including langpacks) a Fedora machine
Also, such measurement would have to be performed on all mirror
server because any single mirror server would provide the data
reliable only to its particular geographical area.
21.08.2018 08:51 Jean-Baptiste <jean-baptiste(a)holcroft.fr> wrote:
Le 17 août 2018 22:06:54 GMT+02:00, Rafal Luzynski
<digitalfreak(a)lingonborough.com> a écrit :
>* do the translations increase the number of Fedora users?
>* are there languages not really used by Fedora users?
The fact that we are asking the first question publicly without raising any
comment worries me.
I think I have formulated the question incorrectly. I should not ask
whether the translations increase the number of Fedora users but
how much do they increase. Or maybe better: how much adding more
languages is helpful. This may be formulated in an optimistic manner:
maybe we have already included all languages people were interested
in and other languages are not interesting for any actual users?
While the answer and collected data will be interesting, the way it
really sounds like business oriented to me.
When I read it, I wonder: do the Fedora community really don't know about the
value of translated content to users?
Well, actually... yes, the Fedora community really does not know.
This question is ambiguous so it needs two answers: of course,
translations are helpful and nobody doubts it. But we don't know
how much are they helpful. Maybe you and me are the only people
interested in high quality translations? Too many times when I asked
people from different countries about their experience with translations
of Fedora and other operating systems, the answer was like "I don't know,
I use English".
I'm confused, because this doesn't reflect the discussion we
had in the past.
Either with our or other i18n contributors.
The research for measuring is something I started and talked about in Boston:
I can see your statistics of the countries where Fedora users
are located and statistics of the translations we provide. That's
useful. But this does not have a chance to measure small minority
languages. Can we measure Upper/Lower Sorbian users in Germany?
Occitan, Corsican, and Basque in France? Hundreds of languages spoken
in India, Russia, and indigenous people of America?
There is almost no way to measure "needs" and bias are
enormous, but there is
some ways about "results" measurements.
I'm unsure we should go in the direction of user tracking like companies, and
go with conviction, as the floss contributor we are.
I'm really far from the idea of tracking the users. But as you can
see this decision has its consequences. It looks like currently we
can't have any way to tell the exact number of languages used by Fedora.