Re: GTetrinet proper UTF-8 support patch

[Vidar Holen <vidar vidarholen net>]

On Wednesday 23 March 2005 23:53, Daniel Carbonell Fraj wrote:
> 1) Extend the Tetrinet protocol to let the server know which encoding is
> using each particular client.
>
> 2) Port the clients to UTF-8 (which shouldn't be hard) and just use
> UTF-8 everywhere.

There is also the hackish option of validating client data as UTF-8, and if it 
doesn't check out, assume the client uses ISO8859-1 (or the encoding du 
jour). Not a permanent solution, but perhaps it could work during a 
transition period? 

> I think 1) is harder to implement, and it can (potentially) break
> compatibility between old and new software. Also, it doesn't fix the
> main issue, which is having a lot of encodings flying around the
> clients, and having to support all of them in each client.
>
And if the client has to be made encoding aware in the first place, it might 
as well convert.

> On the other hand, 2) really fixes the main issue (we'll finally be able
> to forget about encodings) and I don't think it will be harder to
> implement than 1). Also, it almost doesn't require to touch code in the
> server side.
>
> We also have to keep in mind what happens to those clients/servers that
> aren't aware of this issue. They HAVE to be able to play with/against
> UTF-8 enabled clients/servers. Messing with the protocol may render some
> clients/servers incompatible, porting clients to UTF-8 can just render
> some characters unreadable, which isn't a big issue compared to the
> other one.
>
Junk chars are quite tolerable for latin languages which are ascii with the 
occational accent (if one bothers to write them). For cyrillic, not so much, 
and you'd have to resort to ascii interpretations when talking from new 
clients to old (the patch I submittet handles the converse case).

Clients that support cyrillic are probably encoding aware though, so if they 
don't have a switch you can flick, hopefully it wouldn't be a problem for the 
authors to include the little bit of extra functionality. 

> The worst side of all is that we don't have any kind of "Tetrinet High
> Council" or something like that were all the Tetrinet related software
> users and developers can join and discuss these and other matters which
> affect to the free Tetrinet community (tell me if I'm wrong!).
>
> It feels like I'm pushing the rest of the Tetrinet clients developers to
> port their software to UTF-8. Please, do not think so, I just think that
> this is better for everyone, but if the consensus is to leave things as
> they are, I'll do it.