[7021] in SIPB bug reports
Re: "the sipb locker seems to be content..."
daemon@ATHENA.MIT.EDU (Greg Hudson)
Sat Oct 10 09:29:20 1998
To: Salvatore Valente <svalente@MIT.EDU>
Cc: bug-sipb@MIT.EDU
In-Reply-To: Your message of "Sat, 10 Oct 1998 03:24:58 EDT."
<199810100724.DAA14839@StarbaseZero.ne.mediaone.net>
Date: Sat, 10 Oct 1998 09:29:10 EDT
From: Greg Hudson <ghudson@MIT.EDU>
> Of course, vague feelings are useless. Does anyone actually know
> how compatible these systems are? I'd like to see a case made for
> what should and should not be a symlink.
I know of no problems in backwards compatibility (that is, running 5.3
programs on 6.2 or 6.3 systems, or running 6.2 programs on 6.3
systems), and it has been exercised quite a bit. So I think it's
reasonable to point a machine at older binaries if that makes things
easier.
I feel a bit queasy about running 6.3 binaries on 6.2 systems. I
don't know of any specific cases where it doesn't work, and I think
SGI may have gone to some care to make sure that they have pretty much
identical ABIs, but I'm not sure. On the other side, IRIX 6.3 really
is a different version of the operating system from 6.2 (it's not just
a modification of 6.2 which runs on the O2s), and compatibility in
this direction hasn't been exercised much.
You could, of course, just compile everything on 6.2. Hopefully as of
the next Athena release (next summer) we will be running 6.5 on all
SGI machines, so it's not like the Indys are going to go away before
this version skew problem is resolved.
Finally, a question: you made a machtype/sgi6 volume, rather than
reusing the machtype/sgi volume. Does this mean you want to preserve
functionality for IRIX 5.3 machines? Has nothing been recompiled in
the sipb locker yet with any IRIX 6.x version?
