Gordon Matzigkeit <[EMAIL PROTECTED]> writes:
> [I'm writing this message from inside Mindterm 0.97
> (http://www.mindbright.se/mindterm/), a GPLed Java applet that
> implements SSH version 1.x... it's running inside my boss's copy of
> Netscape 4.08.
>
> I'm logged into my home machine, which is running lshd-1999-01-21 plus
> some patches to fix up sshd1 fallback. My home system is now a little
> step closer to depending only on free software. Thank you!]
;)
> After thinking about it a little bit, I believe that it is unnecessary
> to make ssh1-fallback a compile time option. Why should we have two
> seaparate code paths to maintain, when you've already taken the
> trouble to make an object system that handles conflicting requirements
> quite nicely?
I think it would make sense to have a configure time option
--with-sshd1, to make it possible to (i) disable the callback code
completely, and (ii) specify a default path for the sshd1 program. I
agree that it seems unnecessary to try to figure out automatically
whether or not a suitable sshd1 program is available. In any case, a
runtime option is definitely nedded.
> So, without further ado, here are my patches. The first four
> ChangeLog entries are patches to make things work properly (probably
> alredy fixed by you, Niels). The rest are patches to replace the
> --with-sshd1 configure option with a runtime --ssh1-fallback option.
Thanks. Seems there will be a lot of things I have to integrate next
week.
> Next on my agenda: writing a pseudo-terminal object so that Bash
> prints a command prompt and emacs works. This object will use
> different methods for obtaining a pty, depending on the platform. If
> pty support is not known for a given platform, it will use the
> existing make_pipe implementation.
Bazsi is already working on pty support. Please coordinate with him
before you start to work on the pty stuff.
> Have fun, (I sure am!) ;)
Happy hacking,
/Niels
