Re: [translate-pootle] Integrated platform and installers/packages

[Clytie S. <cl...@ri...>]

Sorry it's taken me a few days to answer these...

On 02/10/2009, at 12:01 AM, F Wolff wrote:

> Op Do, 2009-10-01 om 19:26 +0930 skryf Clytie Siddall:
>
> ...
>
>> The question is: are Pootle and Rosetta serving the same audience?
>> If
>> so, how can we be more useful to that audience, now Rosetta has
>> caught
>> up with some of our key advantages?
>
> I have not seen anybody use Rosetta yet, so for the time being I'm
> considering this purely theoretical.  Last time somebody told me, it
> wasn't even easy to get the source code for it.

I don't know if they're still calling it "Rosetta": there's a whole  
thread about this on the mailman-i18n list [1] where they just call it  
"Launchpad translation", but they refer back to my comparisons with  
Pootle and my emphasis on Rosetta/Launchpad not being open source and  
not having access/quality controls: problems they say are now solved.

This means using Rosetta _on_ Launchpad, not installing it separately.
>
>
>> It occurred to me in the night watches that Rosetta's main advantage
>> in this situation is its integration with projects. That's certainly
>> what I have heard, and continue to hear on my project lists. If
>> you're
>> running a project on Launchpad, and you need translations, you don't
>> need to install anything, create any procedures or setup anything
>> much. Rosetta is integrated in Launchpad. Your translation teams can
>> devolve from your development team.
>>
>> I still think the capability to install anywhere is important to
>> Pootle, but I think we need to lower the barrier there (cf. the
>> regular appearance on this list of frustrated and confused posts
>> from
>> people trying to install Pootle). We need to provide installers, and
>> work with the package-management systems.
>
> I guess people who don't struggle to install it just don't write to  
> the
> mailing list :-)  We have a growing list of people who seem to have
> managed quite well to install Pootle:
> http://translate.sourceforge.net/wiki/pootle/live_servers

I'm very sorry if I gave the impression that Pootle has been anything  
less than successful. Many projects and individuals do use it, and I  
support it completely. But when I do see an email to this list asking  
for help with an install which hasn't been straightforward, I wince,  
and wonder for each one of these mails to the list, how many people  
try to install Pootle and give up because it's too hard?

I do think it would help if the Pootle installation process were easier.
>
> As Julen mentioned, Pootle is packaged for several Linux  
> distributions.
> But of course, we should keep working on improving the state of
> packaging where we can.

Yes indeed. Julen's suggestion of packaging for easy_install is  
excellent, since Python is so heavily involved. Amos also stressed  
cooperation between the packagers of the distro, Pootle and the Pootle  
dependencies.

Amos, good to hear from someone on the Squid project!
>
>
>> I also suggest for your discussion the possibility of integrating
>> Pootle with another project-development platform, e.g. SourceForge
>> or
>> GForge.
>> <snip>
>
> I think it is important that we all continue to help and inspire
> software projects to improve things for localisers. Everything you
> mention will definitely be a great help.

I have yet another idea for Pootle itself, which I'll add to the issue  
tracker.
>
> Integration with something like SourceForge or GForge is a great idea,
> and I think we should take it up. I actually think there was already a
> request with SourceForge some time ago.  I hope that the upcoming
> version of Pootle will make all of this much easier.

Version 13 sounds very interesting. What do you have up your combined  
sleeves? :)

Is there any way we can follow up the request to SF? I'll help if I can.

Steve Herrick added:

> This is the lynchpin. It doesn't matter how cool Pootle is if people
> can't use it. I've been trying for a year now to get it up and
> running, and I've failed multiple times. I wouldn't keep trying if I
> hadn't tried everything else first and been unhappy. Make Pootle
> easier to install, and I will sing its praises everywhere I go.

Steve, I'm sorry to hear of your difficulties, but thankyou for adding  
your experience. Pootle is terrific, but if the install doesn't work,  
many people are going to give up.

And I think most devs and translators are just looking for a Pootle  
someone else is hosting.

David Nelson mentioned:

> Would there be any advantage in integrating Pootle with Zikula, the OS
> application framwork/CMS? Since recently, Zikula has been using  
> Pootle at
> http://translate.zikula.org

I have thought about the advantages of combining Pootle with a CMS at  
least. It's definitely a thought. But it will depend on the overall  
direction of Pootle. I'd really like to see it more closely integrated  
with the Translate Wiki: a link to the Wiki's main page or the  
Localization Guide in the Pootle header or footer, and key pages like  
the Plurals and Language Codes available from the My Account page on  
Pootle. I've also been talking on the Decathlon list about how  
Decathlon, the Translate Wiki (or its material) and Pootle could  
combine to form a central resources for translators and developers. [2]

Amos also compared Rosetta and Pootle, since he had used both:

> Rosetta.
>
> Pros:
>  - provide nice fast updates of .po/.pot _out_ of VCS
>  - huge community or translators 'built in'
> Cons:
>  - no VCS commit access.
>  - no record of translators contacts (thus copyright issues)
>  - no controls for moderating specific languages. (ie Algerian has a
> moderator double-checking, but Zulu is free for all edits)
>  - random people translating.
>
>
> Pootle.
>
> Pros:
>  - VCS access
>  - translator personal copyright
> Cons:
>  - less easy to navigate
>  - requires resources to maintain
>  - convincing people to sign up to yet another translation website  
> is a
> hard task.
>
>
> So I'd say the same audience, from polar opposite ends of the field.

According to the Mailman people, Launchpad translation does include  
sync with their bzr branches. That may only apply to Launchpad  
projects: I don't know. It may not include commit access yet.  
Interestingly, one of the "pro" points on the Mailman list was that  
"Launchpad translation" allowed translation suggestions. They do seem  
to have been watching Pootle. ;)

I gather that the level of access/quality control is up to the  
language-team leader and the project leader. You form developer and  
translation teams, and can restrict access to members of the  
translation team for that project. I don't know if they have the sort  
of permissions setup yet that Pootle has.

However, I cannot stress enough the importance of Danilo Segan in this  
comparison. If I had to choose two people I consider consummate  
professionals (however unpaid) from the big FLOSS i18n projects, I  
would choose (1) Christian Perrier and (2) Danilo Segan. And you know  
how good Christian is.

Danilo Segan IIRC created the new translation access/stats interface  
for GNOME ("Damned Lies") [3]. If it weren't to the possible  
disadvantage of Pootle, I'd be breathlessly anticipating what he can  
do at Launchpad, when he's actually paid by Canonical, considering the  
wonders he's achieved at GNOME in bits and pieces of his own time.

from Clytie

Vietnamese Free Software Translation Team
http://vnoss.net/dokuwiki/doku.php?id=projects:l10n

[1] http://www.mail-archive.com/mai...@py.../msg00496.html

[2] http://sourceforge.net/mailarchive/forum.php?thread_name=a699de540910072253me752817w318cf26db1dca6e9%40mail.gmail.com&forum_name=translate-decathlon

[3] http://l10n.gnome.org/languages/
Unfortunately, you need to login to see the best features. It's a very  
good interface indeed, a big improvement on the KDE adaptation used  
before.