*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
Eric,
While you're deciding which distro to get when you jump back on
the upgrade-cycles merrygoround, it may actually be possible
for two versions of glibc to exist on your current system,
allowing you to run programs distributed only as binaries for
current glibc.
For example, on a stock redhat 7.3 system, I can't upgrade glibc
because my entire system depends on the old version. But if
I build a recent glibc (-2.3.2) it installs by default in
/usr/local/lib or /usr/local/glibc-2.3/lib instead of usr/lib,
so doesn't interfere with my current system. Then I can run,
for example lin_ono version 10 by aliasing ono to a command
which sets the .so search path, and starts the new ld_linux
to run my command:
(setenv LD_LIBRARY_PATH \
/usr/local/glibc-2.3/lib:/usr/lib/:/usr/local/lib:/usr/X11R6/lib ; \
/usr/local/glibc-2.3/lib/ld-linux.so.2 /raid/bin/o10/lin_ono)
I've just now got this working so don't yet know how practical or widely
applicable it is. Also I think I had to upgrade gcc and binutils to
the latest version in order to compile the new glibc. I'd be glad for
any advice from the Linux Gurus on this question.
For what its's worth,
Ed
D. Eric Dollins wrote:
~~~
errors. The errors look something like this:
/usr/local/coot/Linux-bubbles-pre-release/bin/coot-real:
/lib/i686/libpthread.so.0: version `GLIBC_2.3.2' not found (required
by /usr/local/coot/Linux-bubbles-pre-release/bin/coot-real).........there
were some more errors looking for /usr/lib/libGL.so.1.
I am running an admittingly old version of Linux redhat 8.0. How can
I get around this dependency issue? I looks through RPMfind.net and
couldn't find any GLIBC_2.3.2 for redhat 8.0. Any idead how to get
around those errors?
Cheers,
Eric
On 10/27/06, Paul Emsley <[EMAIL PROTECTED]> wrote:
On Wed, 2006-10-25 at 16:45 -0400, D. Eric Dollins wrote:
> *** For details on how to be removed from this list visit the ***
> *** CCP4 home page http://www.ccp4.ac.uk ***
>
>
> Hello all,
> I wondered if anyone can help with a NCS problem in coot. I have a
> 3.2A structure with 5 homodimers (10 chains) in the ASU. Given the
> resolution, there are 10 chains for NCS. Obviously, I am disinterested
> in refining all of them. I've tried using NCS in coot, however, when I
> try to apply the edits from my master chain (which I changed from
> chain "A" to the chain with the best density- happens to be "G"), coot
> only applies changes to some of the chains. Not only that, but the
> chains which the changes are applied to also seems to be random.
There indeed was an error in applying NCS edits to other chains. I
thought that I'd fixed that for 0.1.2. Which version are you running?
> I've
> gone through this several times now and the changes have been applied
> to different chains each time. When I load in the model, coot does
> find NCS for all chains A-J.
> I wonder if this might be a buffer issue given the ridiculous size of
> my ASU (>0.5 Megadalton)?
That's not it, I'd guess.
> I noticed also that coot can use a set of strict NCS matrices. Is
> this similar to CNS where you reduce your number of molecules, in my
> case, from 10 to 1. If so, how do you re-apply the matrices to
> re-constiture the full ASU for a run in refmac?
There is no straightforward way in coot to do this. You can mess with
copy-molecule and applying an R/T op to each copy, then merge molecules
(I've not tried to do that). Using pdbset to generate the other
molecules? You'd need to then "concat and edit" the resulting pdb files
- bleugh.
Paul.
