[vlc-devel] [VLC] #3054: Default charset for EPG should be ISO-6937

[Rémi Denis-Courmont]

On Wed, 2 Sep 2009 08:43:46 +0200, Marian Ďurkovič <md at bts.sk> wrote:
>> How about trying ISO_6937 and falling back to ISO_8859-1 if that fails?
>> It seems that most accentuated Latin-1 characters are invalid in
>> ISO_6937 which is a multi-byte encoding.
> 
> I just tested this on one broken channel. Unfortunately, latin-1 texts
> almost never fail iconv when it's tried with iso-6937 first. Here is
> an example what we might get:
> 
> 'Un florilŁge d'extraits musicaux, une faĿon diffØrente et
extrŒmement
> variØe de dØcouvrir la musique, les interprŁtes et les orchestres.'
> 
> Thus we probably need a config option named "Ignore the DVB standard and
> use xxx charset for EPG". This would be similar to VDR approach:
> http://www.linuxtv.org/pipermail/vdr/2008-March/016277.html

We tried it for subtitles. Most users did not find the option, and assumed
"VLC [was] broken". We need to use automated heuristic. There is NO WAY
around.

> BTW, my other patch (Provide charset detection also for SDT fields)
> was still not commited to git, this one is needed to properly decode
> station names if they use accentuated characters.

I'm afraid I lost track of that.

-- 
Rémi Denis-Courmont