> This is only possible if we find a way how to solve the > issues listed in > the mailthread "Oracle: setting the database character set (SET NAMES > not supported)". We can only store the translated strings in > a database > if we can transform all translation to UTF-8. > > Another thing is the performance and translation itself.
Performance is not really a problem. A select statement to get translations needs 0.005 secs. And you only need translations for the webinterface. The log msgs are always in english or? > Nearly all big > Open Source projects are using gettext and I see no reason > why we should > move away from the actual state-of-the-art system. Even KDE with much > more strings than OpenCA uses gettext. What happens if the database > fails? The GNU stuff of gettext is always available. > > Any arguments why we should invest the time in an own system? > Well the reason "because others use it, we use it too" is hmm stupid. Or do jump from a bridge, when nearly all do it? Kde has a huge translation team. They always fix the stuff. So with such a small team, you cant always change all translations. That isnt possible. If the database fails, openca cant operate, so you dont need translations. And the some basic things, like error msg you can still do by gettext. I coded a huge community page for an onlinegame. There i use that translation function. The server is a p3 733mhz. So not a highend machine, but it delivers this complex site with many sql querys in 0.1s German: http://www.mkit.net/?lang=de English: http://www.mkit.net/?lang=en Regards til
smime.p7s
Description: S/MIME cryptographic signature
