[819] in Info-AFS_Redistribution
Re: Compiling ATK5.1 from source.
daemon@ATHENA.MIT.EDU (Christer Bernerus)
Tue Jun 2 05:12:25 1992
Date: Tue, 2 Jun 1992 09:43:22 +0200 (MET DST)
From: Christer Bernerus <bernerus@animal.cs.chalmers.se>
To: Bob_Dew@alw.nih.gov
Cc: Info-AFS@transarc.com, Info-Andrew <info-andrew@andrew.cmu.edu>
In-Reply-To: <ce_WreK0ts4j4KrVI7@alw.nih.gov>
\begindata{text,19768836}
\textdsversion{12}
\template{messages}
Excerpts from info-afs: 1-Jun-92 Re: Compiling ATK5.1 from s..
Bob_Dew@alw.nih.gov (1245*)
I don't believe anything in the Andrew code needs osi_alloc or osi_free,
so what prevents one from removing xrd.o from librx.a before compiling
ATK? We haven't tried this for AFS3.2beta, but Andrew 5.1 builds fine,
using this workaround, under AFS3.1. What's different in AFS3.2beta
that requires the osi routines for an Andrew build?
Your'e probably right about that there's no osi_alloc/free dependency in ATK,
however, Transarc has rewritten librx to use osi_alloc/free, so if I rip out
those routines, the linker complains.
The ATK code itself would probably do fine without the osi routines, but if
they use anything in the rx library that needs them, well ..
In any case, the linker must be kept happy if one wants a clean build.
Chris\
\enddata{text,19768836}
