Re: g_io_channel_read_chars() loop

[Joel Becker <jlbec evilplan org>]

On Thu, Aug 09, 2001 at 09:47:13PM -0700, Ron Steinke wrote:
> Looking back at the Changelog, it looks like the changes I was talking
> about went in on the third, so they're probably already in the
> earlier version you're using (it's the Changelog entry that
> mentions fixing bug 58472).

	Here's what I reverted on giochannel.c.  I actually pulled the diff
on '-D 4 days ago' at the time, so it hit every file that matches this
datespec.

--- giochannel.c        2001/08/03 20:42:06     1.18
+++ giochannel.c        2001/08/05 20:26:09     1.20


> When you say you're using the system encoding, what do you mean?
> If you mean that you're not setting it, the default is always
> UTF-8. If you mean you're setting it to some particular
> system encoding using g_io_channel_set_encoding (), which
> encoding is it?

    ANSI_X3.4-1968 in this particular case.  My code is basically doing:

if (no encoding specified) { set_encoding(g_get_charset()) }

this is based on the assumption that g_get_charset() provides the
encoding that a subprocess is going to use in the current environment.
It has worked just fine up until this particular CVS update.
	Now that you mention it, I am not sure that I tried this with
any other encoding.  It should not have been a problem.  But one must
cover all bases.  The input in these cases looks similar to
"connection_type\nethernet\n", which is very valid ANSI_X3.4-1968.

Joel

-- 

"Up and down that road in our worn out shoes,
 Talking bout good things and singing the blues."

			http://www.jlbec.org/
			jlbec evilplan org