[Localization] New localisation workflow: Overview and status

[Xavier Alvarez]

On Thursday 01 November 2007 16:24, Sayamindu Dasgupta wrote:
SD> Hi,
SD>
SD> On Thu, 2007-11-01 at 11:09 +0100, Marco Pesenti Gritti
SD> wrote: 
SD> > On 11/1/07, Rafael Enrique Ortiz Guerrero <dirakx at gmail.com>
SD> > wrote: 
SD> > > On 10/31/07, Marco Pesenti Gritti <mpgritti at gmail.com>
SD> > > wrote: 
SD> > > > On 10/31/07, Sayamindu Dasgupta <sayamindu at gmail.com>
SD> > > > wrote: 

...snip...
SD> The only reason we wanted submodules was to ensure that Pootle
SD> had write access to _only_ the translations, and nothing else
SD> - but I guess it adds too much complexity for the developers
SD> (and on top of that, migrating existing projects to a
SD> submodule based setup might be a major pain in the
SD> backside :-). I had a small discussion with Xavi on IRC  about
SD> this, and he thinks likewise on the complexity issue. 

At least that's my perception... but I get dizzy with plain git ;)

Security (that Pootle will not mess things up) is a nice feeling, 
but if that comes at the expense of complicating everybody else's 
life is a high price to pay.

Maybe for some developers it would be fairly natural and simple, 
but usually a manual step is a source of unpleasant surprises 
(particularly when this involves L10n and a developer will not 
'test or check' other locales normally or in a systematic way...

"Gee! Why are some strings in foo-lang but not others?"

To the set of possible failure points:
- Is that non-foo string i18n-ized? (ie: the _() trick)
- Did I re-generate the POT with it?
- Is the PO fully translated?
We would now add:
- Did I manually pull the latest submodule?


SD>
SD> I would love to hear some comments from someone from the 
SD> server admin team on this.

Me too! Or anybody else for that matter :)


SD>
SD> Warm regards,
SD> Have a great day,
SD> Sayamindu


Cheers,
Xavier

-- 
XA
=========
Don't Panic!  The Answer is 42