Dear Andreas et al:
My post to the cootbb on this never made it.
I've fixed the coot package to use gpp4, mmdb, ssm, clipper, and
fftw(2) rather than ccp4-onlylibs-dev, which will cut down on the
installation overhead in the future (these packages are minimal).
fftw(3) is not compatible with fftw(2). fftw(2) is "legacy" but
almost everyone in crystallography still seems to use it.
fftw-mpi is a variant of fftw and a superset. It allows the
calculation to be split among multiple processors. For coot it
doesn't matter, but if you already had fftw-mpi installed, coot
shouldn't force installation of the other one (since it is a subset
anyway). Previously I had coot link to static fftw built in ccp4-
onlylibs-dev.
I've also updated the ccp4 package to fix (I hope) compilation and run-
time problems. It built fine for me on 4 different machines.
I've put more details here:
http://tinyurl.com/cootccp4changes
I've also fixed a couple of shlibs packaging errors. As a consequence,
you will probably have to issue the command
fink install coot
TWICE. The second time it goes really quickly (it only unpacks the
debian file).
Similarly for ccp4.
As always, you can grab my debian packages and avoid the overhead of
compiling:
http://tinyurl.com/fink-precompiled
All the best,
Bill
On Sep 12, 2008, at 2:52 AM, Andreas Förster wrote:
Hey there,
running fink update-all, I'm asked for help in picking a package to
satisfy a virtual dependency, with fftw-shlibs and fftw-mpi-shlibs
presented as alternatives. Coot as well as ccp4 have in their list
of dependencies fftw-shlibs | fftw-mpi-shlibs, indicating that I can
go with either. But what's the difference? Why are there two
packages? They're both maintained by the Sebastien Maret, are at the
same version and perform the same task (DFT in one or more
dimensions, of both real and complex data, for arbitrary input size).
Currently on my Leppard system, I have fftw3 and fftw3-shlibs
installed (I didn't follow Kevin's recent posts on the subject), but
not any fftw-related package in version 2. Coot has always compiled
fine, whereas ccp4 never has.
Thanks.
Andreas
--
Andreas Förster, Research Associate
Paul Freemont & Xiaodong Zhang Labs
Department of Biochemistry, Imperial College London
