Re: Terminal parameters patch

Peter Gutmann wrote:
> Stephen Hurd <[email protected]> writes:
>
>> The current SSH client always advertises a terminal type of "xterm" with
>> dimensions of 80x48.  This patch allows control over that (if set before the
>> session is activated).
> The problem with this patch (and an earlier one submitted by Chris Desjardins)
> is that the SSH protocol supports vast numbers of parameters exchanged in
> various messages, which means that every parameter would require a new
> cryptlib attribute and code to handle it, leading to an explosion of
> attributes and code handlers.  In addition since this breaks ABI
> compatibility, it can only be introduced in a minor-version rev (e.g. 3.4.x to
> 3.5.x).  I've been trying to think of ways to handle this without the
> attribute explosion and ABI breakage (thus the long delay in replying), but
> can't think of any easy way to manage this.
>
> If anyone can suggest some generic mechanism for handling this (both in terms
> of cryptlib attributes and in the code) I'd look at incorporating it, but I
> can't really add it in the form of "add lots of new attributes for SSH
> parameters" for the reasons given above.

Hrm... how about the env variables TERM, COLUMNS and LINE as specified
by SUSv2 here:

http://pubs.opengroup.org/onlinepubs/7908799/xbd/envvar.html

TERM would usually already be correct on *nix systems (though I rarely
see COMUMNS or LINES set).  For the server side, it would also
automatically take care of setting the TERM variable for *nix systems
which would also help out.  This obviously wouldn't allow dynamic
updates from the client side, but would be marginally better than the
current hardcoded values.

_______________________________________________
Cryptlib mailing list
[email protected] via Mail: [email protected]
Archive: ftp://ftp.franken.de/pub/crypt/cryptlib/archives/
http://news.gmane.org/gmane.comp.encryption.cryptlib
Posts from non-subscribed addresses are blocked to prevent spam, please
subscribe in order to post messages.