[comp.windows.x] Help!? Maintainence of multiple vendor source trees via NFS.

timothy@spica.ucsc.edu (Timothy Oborne) (08/20/88)

I've been maintaining the X sources here at UCSC for about 6 months.
Until now we only had X running on our Suns. I have source on the main
server  (a 3/280) and NFS mount the include/lib/bin directories on all
of the other machines (mostly 3/50's and 60's).

Recently I have acquired an alpha level server from another vendor of
whose workstations we have maybe 8 or 10.  We have also recently
installed NFS on these same machines.  The point I am driving at is that
I need the benefit of people's experience maintaining multiple source
trees on one machine via NFS.  

The Sun server has lots of disk space (new disks) and the "other"
machine is fairly tight on space.  I wish to consolidate my source trees
as much as possible, as there are few differences aside from the server
code itself.

If you have ANY experience with multiple source trees on the same
machine I would GREATLY benefit from any experience you might care to
relate.  

Please e-mail replies and I will consolidate and re-post to the net if
there is a great enough interest.

Thanks in advance,

Timothy. W Oborne

==============================================================================
Timothy W. Oborne			Internet: timothy@saturn.ucsc.edu
University of California, Santa Cruz	UUCP: ucbvax!ucscc!saturn!timothy
==============================================================================

kek@DINORAH.WUSTL.EDU (Ken Krippner) (08/24/88)

I tried to keep only one source of X and use a shadow tree across NFS.

Unfortunately, I was keeping the source on a micro-VAX and one of the 
shadow trees on Apollo DN3000.  The unfortunate part is that the apollo
has a bug (confirmed by apollo) in lstat so that over the NFS, it gives
unpredictable (to me and to the apollo software as well) response.  I had
to give it up and make another copy on apollo.  (at least 'til SR10)

thought you'd like to know in case these other machines you reference are
apollo.
ken

kek@DINORAH.WUSTL.EDU (Ken Krippner) (08/24/88)

I tried to keep only one source of X and use a shadow tree across NFS.

Unfortunately, I was keeping the source on a micro-VAX and one of the
shadow trees on Apollo DN3000.  The unfortunate part is that the apollo
has a bug (confirmed by apollo) in lstat so that over the NFS, it gives
unpredictable (to me and to the apollo software as well) response.  I had
to give it up and make another copy on apollo.  (at least 'til SR10)

thought you'd like to know in case these other machines you reference are
apollo.
ken

wee@iris.ucdavis.edu (Chris Wee) (08/24/88)

We have all our sources sitting on a VAX 8600 (Ultrix).  There is a shadow
tree on the VAX and another shadow tree on a HP9000s350 (HPUX) via NFS.
We are adding another shadow tree on a SUN somewhere (also via NFS).

I used the MIT distributed lndir.sh script to create the shadow trees
and wrote a little custom script to move some object files around (not
needed any more).  So far, it's been working.  I have generously commented
the lndir.sh script, but haven't had to change it.

Good luck with your site.

lamy@ai.utoronto.ca (Jean-Francois Lamy) (08/25/88)

In article <2907@ucdavis.ucdavis.edu> iris!wee@ucdavis.ucdavis.edu (Chris Wee) writes:
>I used the MIT distributed lndir.sh script to create the shadow trees

On Sun 4s, SunOS 4.0, lndir.sh causes "sh" to dump core.  I kid you not.
The core dumps occur on the
	if [ `(cd somewhere; pwd)` == `pwd` ] 
lines.  I modified lndir to do a where=`pwd` and use $where instead of
the second call, and have encountered no other problems. (we run X.V11R2 on
SunOS4.0 on Sun 2s, 3s, and 4s, but get our source accross NFS from another
group that still runs 3.3).

Jean-Francois Lamy               lamy@ai.utoronto.ca, uunet!ai.utoronto.ca!lamy
AI Group, Department of Computer Science, University of Toronto, Canada M5S 1A4