Right, I've never used Centos but I assume that the Xext library should
exist. It looks like your X11 libraries might be in /usr/lib64 rather
than /usr/X11R6/lib64. (Once upon a time, the X stuff always went into
/usr, but then some bright spark decided that for R6 they should go in
/usr/X11R6, but it seems that some bright spark has now decided it should
all go back to /usr.)
If the X stuff is in /usr/lib64 then check that there is something like
libX11* in there, and then check the same for libXext*. Given the error
message I assume the former exists but not the latter (unless the linker
is checking the libraries in inverse order, for some perverse reason).
In which case this might only be that that library has not been installed.
I hope it's just that this library hasn't been installed or exists
somewhere besides /usr/lib64, e.g. /usr/X11R6/lib64.
On the other hand, if for some obscure reason the Xext library should not
exist on Centos then we need to change the ccpnmr2.0/c/environment.txt
file. You should find a line like:
X11_LIB = -lX11 -lXext
Just change that to
X11_LIB = -lX11
Then in ccpnmr2.0/c type "make" (which might fail still) and (if it does
not fail) "make links". (Unfortunately it's quite possible that the
compilation will "work" but that at run-time you get a complaint that it
can't find certain functions.) You would then need to check whether the
ccpnmr/bin directory was created (with analysis, ..., inside). If so then
you are done. Otherwise you need to run the installCode.py script again
but this time tell it not to do the creation of environment.txt or the
compilation. All we need the script to do this time is to create the bin
directory.
Wayne
On Tue, 18 Nov 2008, Arda Freire, A. wrote:
> Hi Wayne,
>
> Thanks for your fast reply. First of all, I am sorry I didn't mention that I was working from the source code. I think that my problem is that the compilation is not working. The only *.so file I have is named BlockFile.so.
> Also, I have found the next error (/usr/bin/ld: cannot find -lXext) during the installation process in which I didn't realize previously:
>
> Compile and install Analysis code (y or n)? y
> You can create environment file from scratch (this does not copy it from environment_default.txt)
> If you have already created it and edited it by hand you probably do not want to create it again here.
> Create environment file (y or n) (answer y unless you know otherwise)? y
> Does your glut need explicit initialisation (OSX and freeglut probably y, ordinary glut probably n) (y or n)? y
> Use -fPIC compiler flag (y or n) (if gcc compiler y, otherwise n; answer y unless you know otherwise)? y
> Backing up ccpnmr2.0/c/environment.txt to ccpnmr2.0/c/environment.bak
> cd memops/global; make
> make[1]: Entering directory `/opt/ccpnmr/ccpnmr2.0/c/memops/global'
> cc -shared -o GlHandler.so -L/usr/lib64 -L/usr/lib64 -L/usr/lib64 -L/usr/lib64 \
> utility.o python_util.o gl_handler.o py_gl_handler.o py_tk_util.o w32func.o -lglut -lGLU -lGL -ltk8.4 -ltcl8.4 -lX11 -lXext -lm
> /usr/bin/ld: cannot find -lXext
> collect2: ld returned 1 exit status
> make[1]: *** [GlHandler.so] Error 1
> make[1]: Leaving directory `/opt/ccpnmr/ccpnmr2.0/c/memops/global'
> make: *** [global] Error 2
> Do you want to create a bin directory (answer y unless you know otherwise) (y or n)? y
> Creating the bin directory
> Install latest updates (from server) (y or n)? n
> Run Analysis (as test) (y or n)? y
> Model read finished. Duration 2.34465408325
> Model validity check skipped
> Error, the Analysis module will not work, something is wrong with the C code.
> Traceback (most recent call last):
> File "/opt/ccpnmr/ccpnmr2.0/python/ccpnmr/analysis/AnalysisGui.py", line 71, in ?
> from ccpnmr.analysis.AnalysisPopup import AnalysisPopup
> File "/opt/ccpnmr/ccpnmr2.0/python/ccpnmr/analysis/AnalysisPopup.py", line 87, in ?
> from ccpnmr.analysis.Analysis import Analysis
> File "/opt/ccpnmr/ccpnmr2.0/python/ccpnmr/analysis/Analysis.py", line 85, in ?
> from memops.c.MemCache import MemCache
> ImportError: No module named MemCache
>
>
> I think the libraries are properly installed... Do you have any suggestion for this?
>
> Best regards,
> Ana
>
>
> -----Original Message-----
> From: CcpNmr software mailing list on behalf of Wayne Boucher
> Sent: Tue 11/18/2008 5:11 PM
> To: [log in to unmask]
> Subject: Re: Error during Analysis v.2 installation
>
> Hello,
>
> Did you install from source code or from a pre-compiled version? I'm
> first going to assume the former.
>
> So if you installed from source code then it sounds suspiciously like
> either the code wasn't compiled or the symbolic links weren't set up.
>
> The C code is all under the /opt/ccpnmr/ccpnmr2.0/c directory. So, for
> example, if you look in /opt/ccpnmr/ccpnmr2.0/c/memops/global, do you see
> any *.so files? If so, the code probably compiled ok. In particular
> check for MemCache.so, because that's what the error message below is
> complaining about.
>
> If that is ok then next check the symbolic links. We use symbolic links
> from the python to the C directories because Mr Python completely goofed
> on how imports are done in Python. So if you look in
> /opt/ccpnmr/ccpnmr2.0/python/memops/c do you see any symbolic links, e.g.
> from MemCache.so to /opt/ccpnmr/ccpnmr2.0/c/memops/global? If not then
> you can do one of two things:
>
> (1) cd /opt/ccpnmr/ccpnmr2.0/c
> make links
> (note that this should have happened during installation, so if it didn't
> then this step might fail here as well)
>
> *OR*
>
> (2) cd /opt/ccpnmr/ccpnmr2.0/python/memops/c
> source linkSharedObjs
> cd /opt/ccpnmr/ccpnmr2.0/python/ccp/c
> source linkSharedObjs
> cd /opt/ccpnmr/ccpnmr2.0/python/ccpnmr/c
> source linkSharedObjs
>
> Then hopefully Analysis will work.
>
> If you installed Analysis from a pre-compiled version then it's possible
> there is some incompatibility from what we compiled to what your computer
> accepts. But normally I would have expected a different error message
> than the one you got.
>
> Wayne
>
> On Tue, 18 Nov 2008, Arda Freire, A. wrote:
>
> > Hi,
> >
> > I was installing on a Centos 5, Analysis v.2. It seemed to be working fine but during the test I got the next error:
> >
> > Run Analysis (as test) (y or n)? y
> > Model read finished. Duration 1.8960621357
> > Model validity check skipped
> > Error, the Analysis module will not work, something is wrong with the C code.
> > Traceback (most recent call last):
> > File "/opt/ccpnmr/ccpnmr2.0/python/ccpnmr/analysis/AnalysisGui.py", line 71, in ?
> > from ccpnmr.analysis.AnalysisPopup import AnalysisPopup
> > File "/opt/ccpnmr/ccpnmr2.0/python/ccpnmr/analysis/AnalysisPopup.py", line 87, in ?
> > from ccpnmr.analysis.Analysis import Analysis
> > File "/opt/ccpnmr/ccpnmr2.0/python/ccpnmr/analysis/Analysis.py", line 85, in ?
> > from memops.c.MemCache import MemCache
> > ImportError: No module named MemCache
> >
> > Do you have an idea about which is the problem? Any suggestions?
> >
> > Best regards,
> > Ana
> >
>
|