ssh argv[0] support

[mouring at etoh.eviladmin.org]

On Fri, 14 Sep 2001, Jonathan Amery wrote:

> In article <Pine.BSO.4.33.0109131415190.22017-100000 at etoh.eviladmin.org> you write:
> >
> >I would rather see people submit C code to correct the issue than
> >shell script hacks.  Nothing personal, but lets solve this problem
> >correctly and not half-ass.
> >
>  In what way is this (POSIX shell) script a 'hack'?  It's already been
> decided that it's impossible to do it right in the main ssh code (what if
> you rename ssh was IIRC the main argument), and C is not the right
> language to do the job (particularly given it's dodgy string
> handling).
>
No it was decided that more care should be taken while doing it and not
adding in a massive function from another program which could have
security issues because of it's #ifdef complexity (the one suggested was
from sendmail).

>  Also OpenSSH already distribute one shell script (in contrib) -
> ssh-copy-id (which *is* dodgy and hacky, but I'm working on it).
>
I don't see how this related to setproctitle() emulation.  ssh-copy-id is
a script that that copies your *.pub file to another location and sets it
up in authorized_keys.  I don't see the connection.

>  And I might note that one comment in the original discussion was that
> a shell script was precisely the right way to do it.  If you don't
> want 'hacky' shell-scripts then feel free to revert to the old way of
> doing it, or write an equivilantly portable and bug-free C-programme.
>
What converstation?  Last I remember on the list Steve stated that we
should take one OS at a time to assure it was correct.  HP/UX is done.  It
works native on those OSes with setproctitle().  Just leaves OSes like
Linux, NeXT, etc to look at.

- Ben