Having it installed yourself is one option. If you still prefer
Fink's version, the following thread may be of use to you.
Begin forwarded message:
> From: Murali Vadivelu <[log in to unmask]>
> Date: 15 January 2007 15:44:39 GMT
> To: [log in to unmask]
> Subject: Re: Python 2.5 error
> Reply-To: CcpNmr software mailing list <[log in to unmask]>
>
> Those scripts are available only in the fink installed version of
> CcpNmr. If you have installed CcpNmr yourself, you have probably
> installed it in a folder where you have write permission, unlike in
> Fink where only 'root' has write permission. If it is a folder that
> is write enabled for the user, the python files get byte-compiled
> automatically, the first time they are used.
>
> Hence, you need to byte compile using the script only for fink
> installed CcpNmr and/or other installations of CcpNmr where the
> user do not have write permission.
>
> You can find it in /sw(or whatever your fink folder is called)/bin/
> folder and it is fink specific. You can modify it for any
> installation though.
>
> On 15 Jan 2007, at 15:32, Johnny Croy wrote:
>
>> Murali,
>>
>> Thanks for the update. After uninstalling the fink ccpnmr and
>> installing the one from the ccpnmr website, the speed issues have
>> not been as much as a problem. Also thanks for the scripts to run
>> after updating (which I recently did). However, I can't seem to
>> find them in the ccpnmr directory where I installed. How exactly
>> do you run these scripts?
>>
>> Thanks again for your help.
>>
>> -J
>>
>> On Jan 13, 2007, at 2:25 PM, Murali Vadivelu wrote:
>>
>>> I had a think about it the speed issues and fink installed CcpNmr.
>>>
>>> Assuming you weer using the same python version of CcpNmr (ccpnmr-
>>> py25), the speed issue could be down to having your .py files not
>>> byte-compiled. This is often the case because the fink directory
>>> is only root enabled. Once you update the .py files and run the
>>> program as a non-root user, it would not get byte-compiled, and
>>> this would be the case till that particular file is used with
>>> root privileges.
>>>
>>> Hence, fink installed CcpNmr comes with a few scripts that you
>>> have to run like any usual command from the command line; run
>>> ccpnmr-bytecompile (after any of the .py files are update from
>>> within analysis) & ccpnmr-postupdate (after any of the .c and/
>>> or .py files are update from within analysis)
>>>
>>> Hope this solves the problem. Please keep me updated on this.
>>>
>>>
>>> On 12 Jan 2007, at 19:37, Johnny Croy wrote:
>>>
>>>> Hello again.
>>>>
>>>> I posted earlier about CCPnmr running slow on my powerbook, but
>>>> found that
>>>> completely removing fink and reinstalling it helped tons in
>>>> terms of speed.
>>>> Instead of using the fink version of analysis, the package
>>>> downloaded from
>>>> the CCPNMR site is much better (speed wise).
>>>>
>>>> I am having one issue that I don't know how to fix. After
>>>> staring analysis
>>>> I get a number of errors that read,
>>>>
>>>> python2.5(6235) malloc: *** Deallocation of a pointer not
>>>> malloced:
>>>> 0x50db70; This could be a double free(), or free() called with
>>>> the middle of
>>>> an allocated block; Try setting environment variable MallocHelp
>>>> to see tools
>>>> to help debug
>>>>
>>>> Any ideas of what may be causing this? It apparently doesn't
>>>> have any
>>>> significant consequences when running analysis (at least yet!),
>>>> so I will
>>>> let you all know if I run into further problems. For reference,
>>>> I am using
>>>> the current analysis package with the fink compiled tcl/tk,
>>>> python2.5 and
>>>> mesa packages.
>>>>
>>>> -J
On 7 Feb 2007, at 20:34, Johnny Croy wrote:
> Jo,
>
> I ran into some of the same problems that you are describing. I
> was running the FINK installed version of analysis and working
> within was very painful. I tried the same suggestions as given by
> the fellow CCPNMR users, but still found that the performance
> issues were not acceptable.
>
> In the end I ended up dumping the FINK version and directly
> compiling it using the version that is downloadable from the CCPNMR
> site directly. Performance was noticibly increased and the
> "slowness" that I found to be prohibitive in the FINK version were
> less noticble in the CCPNMR version. However, I have found that
> the implementation of reading in a contour file really, really
> helps the performance issues (as expected and pointed out by Tim
> and Wayne). Additionally, downloading the CCPNMR version and
> compiling in say your home directory allows for updates to be
> easily implemented, as FINK installs in /sw and requires root/sudo
> privileges for upgrading which I found to be a nuisance.
>
> However, with that being said I still think that performance is
> still an issue on the mac that I use (G4 powerbook, 1.5MHz 1.5 GB
> RAM). Since I only have ANSIG to compare to, the graphics
> (redrawing rates, etc) pale in analysis pale in comparison to what
> I can get from ANSIG. However, the ability to switch "on-the-fly"
> from precontoured to "contouring-on-the-fly" is very handy and is a
> signifcant over the ANSIG situation. Try to compile the CCPNMR
> version that you can download from the main site and see if this
> helps! Please let me know if you run into any problems and I may
> be able to help.
>
> Good luck!
>
> - Johnny
>
> On Feb 6, 2007, at 5:58 PM, Jo Claridge wrote:
>
>> Hi
>>
>> Using Fink have installed analysis on a core duo 2 2.33 GHz
>> powerbook pro
>> with 2 Gb of RAM. The programme seems to work OK except that when
>> I load
>> .ucsf or Bruker format 3D spectra the update is prohibitively slow
>> and when
>> looking at more than one plane the programme often crashes. I
>> appreciate
>> that the spectral files are large (up to 500Mb) so I was wondering
>> whether
>> this is to be expected or whether there is anything I can do to
>> fix this. I
>> have used both the openGL and tcl/tk. Do you have any plans to
>> speed up the
>> graphics handling?
>>
>> Jo Claridge
>>
>> IFS
>> Massey University
>> Palmerston North
>> New Zealand
>
> ------------------------------------------------------------
> Johnny E. Croy, Ph.D.
> NIH Post-Doctoral Fellow
> University of Colorado, Boulder
> Wuttke Lab
> Campus Box 215
> Cristol Chemistry
> Boulder, CO 80309-0215
>
> Lab Phone: (303) 492-2369
> Lab Fax: (303) 492-4576
>
> ------------------------------------------------------------
>
>
|