Re: UTF8

[nisse%lysator.liu.se@localhost (Niels Möller)]

markus%gyger.org@localhost (Markus Gyger) writes:

> In the long term, is the idea to use UTF-8 or to have it
> configurable?

It seems that the world is slowly moving towards utf-8, but that
development is not really in scope of the secsh wg to decide. We just
go with the flow.

When it has been necessary to deal with different character sets in
the ssh protocols, so far the answer has been to require utf-8 on the
wire.

> Since currently only US-ASCII is used it's not
> much of a problem

Huh? I'm using latin-1 over ssh sessions all the time. I'm sure others
are using utf-8 as well as a bunch of other encodings.

> But how about e.g. the names of environment
> variables to be passed or the TERM names passed -- should
> they use the same encoding as the channel data (once they
> are outside of US-ASCII) or UTF-8 per default?

For names and values of environment variables, it's about as undefined
as the charset for channel data. For the TERM names, these are fairly
standardized, and ascii. We don't need to care until non-ascii
terminal names are added to termcap/terminfo, and I doubt that will
ever happen.

Using plain ascii for various identifiers which are intended to be
interpreted by machines rather than humans, seems to be a very deeply
rooted convention.

/Niels