Hello,
It is supposed to set CCPNMR_BIN_DIR to the bin/ is and the CCPNMR_TOP_DIR
to one level up, i.e. to the ccpnmr/ directory. But it looks like the way
it is doing this in csh is broken. Dan is looking into it. The bash
version seems to work ok. The bash version is:
export CCPNMR_TOP_DIR="$(dirname $(cd $(dirname "$0"); pwd))"
That did not work in csh. But it looks like the csh variant we used does
not work in many circumstances. It seems to work, for example, if you run
the script from the ccpnmr directory but not from the ccpnmr/bin
directory.
The basic idea is to try and figure out where the script is located as an
absolute path so that we can then figure out where the ccpnmr/ directory
is. You would think that Unix people might have had a command for this
since it's a totally trivially obvious thing you might want to do, but no,
they haven't. At least the bash people seem to have a way of
accomplishing this with some hack, but the csh is proving a bigger
problem. Dan will sort it out (unless someone out there knows how to do
it).
Wayne
On Tue, 10 Aug 2010, Christoph Brockmann wrote:
> Hi Everyone,
>
> I've just upgraded an 64Bit precompiled release and find myself unable to
> start analysis again.
>
> it seems that the things went worng in the startup-script (analysis2)
> because:
>
> setenv CCPNMR_BIN_DIR "$cwd/`dirname "$0"`"
> setenv CCPNMR_TOP_DIR "`dirname "${CCPNMR_BIN_DIR}"`"
>
> set both the TOP and the BIN to wherever the startup-script is located.
> Meaning that everything after that fails for obvious reasons.
>
> Christoph
>
>
> On 08/10/2010 11:21 AM, Wayne Boucher wrote:
>> Hello,
>>
>> If you are running a source code release then Dan has figured out why the
>> upgraded to 2.1.4 has not coped correctly. So after the compilation in the
>> c/ directory the upgrade script has to make symbolic links in the python/
>> directory. Only the normal way of doing that outside of a script ("make
>> links") does not work inside that script for some reason I forget (at least
>> some people found it didn't work) so instead it goes into the individual
>> subdirectories and runs some other scripts. Well the relevant subdirectory
>> is not listed for the BayesPeakSeparator. This means that another way to
>> fix this is to:
>>
>>> cd ccpnmr/ccpnmr2.1/c
>>> make links
>>
>> And then 2.1.4 should be ok. Note, this is only for source code updates.
>> If you have a pre-compiled update and are hitting a problem with
>> BayesPeakSeparator then something else must be going wrong.
>>
>> Wayne
>>
>> On Tue, 10 Aug 2010, Wayne Boucher wrote:
>>
>>> Unfortunately we look set to have a 2.1.5 pretty damn soon (because it
>>> turns out there is a small stupidity in the C code for printing windows),
>>> but I'd like to understand this issue about the BayesPeakSeparator code
>>> before we try and produce a 2.1.5.
>>>
>
|