Re: bug ?

[Andrew Suffield <asuffield@debian.org>]

On Sun, Apr 07, 2002 at 12:38:16AM -0500, Dan Udey wrote:
> > No, I am saying that the definition of "same case" that xchat uses
> > could be altered without breaking anything, and in fact fixing a
> > breakage that can be seen on dalnet.  Exactly how could it cause
> > breakage on RFC-compliant networks?  How do you think other IRC
> > clients do it?
> 
> Poorly.
> 
> IRC servers pass on what other people say. If I send \foo\, it sends
> \foo\. If I send |foo| it sends |foo|. If I join |foo|, and an oper kicks
> me out of \foo\, should my client treat these differently? You're saying
> yes, even though to anyone else (and to the IRC server), I've been kicked
> out of the channel in question.

The server should send you the canonical name of the channel
regardless (many of them don't, annoyingly). There isn't a
particularly good solution to this.

> As a side note, OPN's IRC daemon seems to break as well. Observe:
> 
> [ /join #\foo\ ]
> [ /join #|foo| ]
> ( XChat's #\foo\ tab changes to #|foo| )
> [ /whois Darien ]
> 
> Peculiar. Andrew Suffield, are you still around? Any explanation from OPN
> programmers?

config.h.opn:
/* SANE_CASE_CONVERSION
 * This causes {, }, |, and ~ to be considered different characters to [, ], \, and ^
 */
#define SANE_CASE_CONVERSION

I refer you to the POSIX standards on case matching in the C locale.

This was done partly because the rfc1459 definition is stupid, and
partly because the ~ character was reserved for another purpose (which
will hopefully be completed and deployed in the near future), and
justified by the behaviour of bahamut (since it doesn't seem to cause
dalnet much of a problem).

I guess an entry in the 005 numeric for this is possible, but given
the behaviour of ircnet on that one, the chances of getting everything
changed aren't good; certainly I'm not going to waste time trying to
talk sense into bahamut coders. If somebody can get dalnet to add a
flag to their numeric, I'll include it in dancer for the sake of
compatibility (but OPN's interest in maintaining compliance with
rfc1459 is minimal; compatibility with rfc1459-like clients will be
maintained to a reasonable level, but the intent is to move towards a
saner protocol).

Meanwhile, might I suggest a config option in xchat? (preferably
per-connection, if that's possible)

-- 
  .''`.  ** Debian GNU/Linux ** | Andrew Suffield
 : :' :  http://www.debian.org/ | Dept. of Computing,
 `. `'                          | Imperial College,
   `-             -><-          | London, UK

PGP signature