search

PiRSquared17 / cing

Automatically exported from code.google.com/p/cing
0 stars 0 forks source link

Numpy version needs to be updated #239

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago 
What steps will reproduce the problem?
1. Reading a pickle with numpy.dtype data. E.g. when unit testing: 
cing --test

testMoleculePlot (cing.Libs.test.test_NTMoleculePlot.AllChecks) ... ERROR

===========================================================
===========
ERROR: testMoleculePlot (cing.Libs.test.test_NTMoleculePlot.AllChecks)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Users/jd/workspace35/cingStable/python/cing/Libs/test/test_NTMoleculePlot.py", line 
105, in testMoleculePlot
    project.validate(parseOnly=False, htmlOnly=True, doProcheck=False, doWhatif=False, 
doWattos=False, doTalos=False)
  File "/Users/jd/workspace35/cingStable/python/cing/core/classes.py", line 985, in validate
    doWattos = doWattos, doTalos = doTalos)
  File "/Users/jd/workspace35/cingStable/python/cing/core/validate.py", line 137, in validate
    project.runCingChecks(toFile=True)
  File "/Users/jd/workspace35/cingStable/python/cing/core/classes.py", line 988, in 
runCingChecks
    return runCingChecks(self, toFile=toFile)
  File "/Users/jd/workspace35/cingStable/python/cing/core/validate.py", line 97, in 
runCingChecks
    project.validateDihedralCombinations()
  File "/Users/jd/workspace35/cingStable/python/cing/core/classes.py", line 962, in 
validateDihedralCombinations
    return validateDihedralCombinations(self)
  File "/Users/jd/workspace35/cingStable/python/cing/core/validate.py", line 1481, in 
validateDihedralCombinations
    hPlot.initHist()
  File "/Users/jd/workspace35/cingStable/python/cing/Libs/html.py", line 312, in initHist
    dbaseTemp = cPickle.load(dbase_file)
ValueError: can't handle version -1 of numpy.dtype pickle

I'm trying to update now from 1.3 on production to:
  py26-numpy @1.4.0_0 (active)
sudo port upgrade py26-numpy --force

Original issue reported on code.google.com by jurge...@gmail.com on 29 Apr 2010 at 4:14

GoogleCodeExporter commented 9 years ago 
Do not execute the --force with the upgrade it will take 2 hours.

Instead do something like this:
sudo port uninstall py26-matplotlib py26-numpy
sudo port install py26-matplotlib py26-numpy

This should take less than one hour. I did have to babysit it through some 
further complications but they are 
probably unique to nmr's production setup.

Ideally I wouldn't pickle specific numpy data structures but for now I have no 
time to recode/rerun all.
I also found one difference in the pickling but I would have to test to see if 
it matters really.
This workaround is verified with unit testing and iCing for python setup.

Original comment by jurge...@gmail.com on 29 Apr 2010 at 8:19

GoogleCodeExporter commented 9 years ago 
On my Ubuntu VM numpy is also still 1.3.0:

print numpy.version.version

It would mess up my aptitude to install this manually.

Original comment by jurge...@gmail.com on 31 May 2010 at 1:31

GoogleCodeExporter commented 9 years ago 
I'm updating to Ubuntu 10.04 LTS to overcome the numpy problem.

Original comment by jurge...@gmail.com on 4 Jun 2010 at 8:09

GoogleCodeExporter commented 9 years ago 
Nice, the new version still only has 1.3.0. Time to debug this for real.

Original comment by jurge...@gmail.com on 4 Jun 2010 at 9:48

GoogleCodeExporter commented 9 years ago 
Fixed in r789 by cPickle writing in protocol 2 by explicitly using the number 2 
instead of -1 which should have 
selected the same.

Checked fine on Ubuntu 10.

Original comment by jurge...@gmail.com on 4 Jun 2010 at 1:44