Re: PostgreSQL website translations

Adrian Maier a écrit :
> On Mon, Mar 10, 2008 at 11:49 AM, Magnus Hagander <magnus(at)hagander(dot)net> wrote:
>> On Mon, Mar 10, 2008 at 11:32:28AM +0200, Adrian Maier wrote:
>> > On Sun, Mar 9, 2008 at 1:00 PM, Magnus Hagander <magnus(at)hagander(dot)net> wrote:
>>[...]
>> > Perhaps it would be useful to have the things that need to be fixed
>> > added on Trac ?
>>
>> That's a very good idea. First we just have to agree on what they are :-)
>>
>> I think one big question that we never did solve the last time is what to
>> do with pages that are out-of-date. E.g. when the english version of a page
>> is updated, and the translated one isn't, what do we do with it... Show it
>> anyway, remove it, somehow indicate it... Once we've decided what we want
>> to do about it, then we can start debating how to do it (which is where the
>> point of cvs/svn etc comes back in), and start adding tickets for them.
>
> When an English page is modified we could :
> - immediately send a notice to the translators. Automatically added Trac
> tickets seem to be a cool idea if it's possible.
> - keep displaying the old translated page for 4-5 days

Nope. If a page is modified, it should just display the english one.

> - if the translation is not yet ready we could invalidate the
> page : move it somewhere else and start displaying the
> English page until the translation is ready.
> Or, instead of moving away the page we could add an
> "outdated page since yyyy-mm-dd" warning .
>

I don't think we need to move the obsolete translated webpage. The PHP
code should simply display the english one.

> Either way , I think that it would be better to give the translators
> a few days to do the update before hiding or marking the page.
>

Nope. There's no reason to do this. If the english page fixes an error
(content, not grammar or things like this), I prefer to have the fixed
english one rather than the translated but wrong one.

> But first of all, we'll need a good way to store the last_modification_date
> for each page : each page (English and translations alike) need to
> be "stamped" with a date or date+time . The date of the svn
> revision is not good enough because relying on it is the perfect
> recipe for missing modifications if they are occuring before the
> translators finishes the translation of a previous version.
>
> Here is a scenario :
> - 8-apr: the English page is modified+committed
> - 8-apr: the translator starts translating
> - 9-apr: the English page is modified+committed again
> - 10-apr: the translation gets committed (but it has the contents for 8-apr)
>

Oh OK. I understand your example. But if we can't rely on a svn hook to
stamp each file, I also don't think we can rely on a manual
modification, do we ?

> So, I think that we'll need to build some mechanisms for dealing
> with the page versions/trabslations outside of what does svn offer.
>

Do you have any ideas of what this could be ?

There's another issue that we'll need to answer : localised search. If I
remember well Magnus's talk at FOSDEM, the search on www.postgresql.org
relies on an english only dictionnary. How will french, romanian or
spanish pages affect the current search feature ?

Regards.

--
Guillaume.
http://www.postgresqlfr.org
http://dalibo.com