On Thu, Nov 3, 2011 at 9:59 AM, Jan-Philip Gehrcke
<jgehrcke.googlemail.com>wrote:
> Huhu,
>
> there are questions left:
>
> - was it "expected" that AmberTools/netcdf build failed with Amber
> already being in the same directory tree?
>
Definitely not. While I typically build the development version of the
software rather than the release versions, I have built AmberTools 1.5 --
Amber11 on numerous occasions on numerous platforms, and I've never had
this issue. (In fact, the process was intended to be general enough that
AmberTools 1.5 could be extracted on top of AmberTools 1.4, replace it, and
build fine, but I don't know if it ever worked out that way).
The AmberTools build process should be completely independent of the
existence of, or the state of, the Amber files that may be in the AMBERHOME
directory alongside it (which makes AmberTools a standalone package).
> - if it was not expected: why did it fail for me / should I further
> investigate?
>
Because of *some* left-over or corrupted file in your NetCDF directory, or
a residual incompatible file from a "different" compiler build that "clean"
may not have gotten rid of (if you ran clean, that is)... Short story:
there are way too many files that get created, compiled, destroyed, etc. to
actually debug what happened (and now that information is gone unless you
saved the state of your tree when it happened). I bet you would have a
hard time reproducing that issue now starting just with the two tarballs.
(Try running Dan's sequence, even exchanging steps 3 and 4, which I usually
do in reverse order myself).
All the best,
Jason
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Nov 03 2011 - 08:30:04 PDT
