[GCompris-devel] Proposal for detection of audio resource set

[Bruno Coudoin]

Le 31/07/2015 17:02, Holger Kaelberer a écrit :
> Hi
>
> On 30.07.2015 21:15, Bruno Coudoin wrote:
>>
>>
>> Le 30/07/2015 21:06, JAZEIX Johnny a écrit :
>>> Hi,
>>>
>>> I agree with the fact that rebuilding every time we have a special
>>> locale case is not a good thing.
>>>
>>> If I understand, the 'Contents' file is downloaded at beginning but is
>>> it updated somehow for users (automatically or manually) if it is
>>> updated on server?
>> Yes, we download it each time we run GCompris and save it locally in the
>> cache dir.
>>
>>>
>>> If it's automatic, I don't see any issue with using the file and it
>>> will be easier when adding new voices.
>>>
>> I don't see any issues either but let's wait for Holger's advice, he
>> knows this part very well.
>
> This should be possible. You'd need to change the download logic and
> DownloadManager's public interface a bit. The starting point (e.g. in
> updateResource()) by now is a rcc filename. In order to determine the
> filename, following your propsal, you'd need to download Contents
> *first*.
>
> Of course only if checkDownloadRestriction() == true.
>
> I don't know how your full-rcc build works with respect to the
> Contents file, it should be packaged then.
>
> On the other hand: How often do you expect new country specific
> locales to be added? With zh_TW we'll have three such exceptions. ...
> Just to weigh the efforts ... ;-)

Hi,

Looked at it and I am obsessed by the fact that the content file knows
precisely what voice resource we have or not and that we don't use this
information.

But I agree that it makes too many changes and is error prone. I just
added zh_TW in our hardcoded long locale exceptions.

Bruno.