Re: [JDEV] charsets (was: Protocol extension?)

[patrick <patrick@kia.net>]

> Can I also suggest that there is a marking for language as well. I
> think that it is quite important to provide as much information as
> possbile. If you provide a language tag then I can interface my client
> to an automatic translation system and get messages translated without
> having either to tell the system the language by hand or else trying to
> guess the language (which is hard). There are well defined tags for
> languages that can just be picked up. I think that language and encoding
> are essential additions to the coding.
> 

	You beat me to the suggestion by only a few minutes. If language
isn't provided for in the base protocol, we'll just have to layer it on 
later.

	For purposes of simplicity, and easily implemented clients, these
message attributes should be optionally included by clients, with clients
expected to default sensibly (say, UTF-8, support for which is required by
XML parsers, and US English).  If clients choose to ignore other languages
and charsets, there's not much to be done about it, but we can encourage
them by putting support directly in the protocol.

	The language tags would also be useful to text to speech
synthesizers.


	Speaking of the protocol, anyone know why errors are implemented
as messages? It seems more logical to me to have a separate node. I expect
I'm missing something.


Patrick




> > Sorry to jump on you like this. Most European countries have some
> > national characters not enclosed in us-ascii and it is frustrating
> > having those issues continually marginalised by US developers. I can
> > only imagine how the Japanese feel about it.
> 
> Oh, the Japanese really, really hate it and so do the Russians and the
> Chinese, and the Koreans, actually *everybody* hates it apart from the
> people in the USA...
> 
> L.
> -- 
> http://catless.ncl.ac.uk/Lindsay
> 
>