Comparing the output of pip in the magical venv that can run the alexnet_based and a fresh install that can't. Have saved both of these to different text files:


In [1]:
cd ..


/afs/inf.ed.ac.uk/user/s08/s0805516/repos/neukrill-net-work

In [2]:
!cat magical.freeze


Jinja2==2.7.3
MarkupSafe==0.23
Pillow==2.7.0
PyYAML==3.11
Pygments==2.0.2
Theano==0.6.0
argparse==1.3.0
backports.ssl-match-hostname==3.4.0.2
certifi==14.05.14
distribute==0.6.34
holoviews==0.7
ipython==3.0.0-rc1
jsonschema==2.4.0
matplotlib==1.4.3
mistune==0.5
mock==1.0.1
-e git+git@github.com:Neuroglycerin/neukrill-net-tools.git@7b7eef9c3f9b5ba681f6d3ca55a4b457bc325a37#egg=neukrill_net-dev
nose==1.3.4
numpy==1.9.1
param==1.2.1
ptyprocess==0.4
py==1.4.26
-e git+git@github.com:lisa-lab/pylearn2.git@9a04142875779077b1c276c3e5db7de8c8257773#egg=pylearn2-dev
pyparsing==2.0.3
pytest==2.6.4
python-dateutil==2.4.0
pytz==2014.10
pyzmq==14.5.0
scikit-image==0.10.1
scikit-learn==0.15.2
scipy==0.14.0
six==1.8.0
terminado==0.4
tornado==4.1
wsgiref==0.1.2

In [3]:
!cat fresh.freeze


Jinja2==2.7.3
MarkupSafe==0.23
Pillow==2.7.0
PyYAML==3.11
Pygments==2.0.2
Theano==0.6.0
argparse==1.3.0
backports.ssl-match-hostname==3.4.0.2
certifi==14.05.14
distribute==0.6.34
ipython==3.0.0-rc1
jsonschema==2.4.0
mistune==0.5
-e git+git@github.com:Neuroglycerin/neukrill-net-tools.git@7b7eef9c3f9b5ba681f6d3ca55a4b457bc325a37#egg=neukrill_net-dev
nose==1.3.4
numpy==1.9.1
ptyprocess==0.4
py==1.4.26
-e git+git@github.com:lisa-lab/pylearn2.git@9a04142875779077b1c276c3e5db7de8c8257773#egg=pylearn2-dev
pytest==2.6.4
pyzmq==14.5.0
scikit-image==0.10.1
scikit-learn==0.15.2
scipy==0.14.0
six==1.8.0
terminado==0.5
tornado==4.1
wsgiref==0.1.2

Comparing line by line by eye is a bit annoying, so using Python:


In [4]:
magical = []
with open("magical.freeze") as f:
    for l in f:
        magical.append(l)
fresh = []
with open("fresh.freeze") as f:
    for l in f:
        fresh.append(l)

Then we can just use sets to compare:


In [5]:
set(magical) - set(fresh)


Out[5]:
{'holoviews==0.7\n',
 'matplotlib==1.4.3\n',
 'mock==1.0.1\n',
 'param==1.2.1\n',
 'pyparsing==2.0.3\n',
 'python-dateutil==2.4.0\n',
 'pytz==2014.10\n',
 'terminado==0.4\n'}

Installing just those in the fresh and trying again. Installing in this order:

  • installed mock: errors

In [7]:
fresh = []
with open("fresh.freeze") as f:
    for l in f:
        fresh.append(l)
set(magical) - set(fresh)


Out[7]:
{'holoviews==0.7\n',
 'matplotlib==1.4.3\n',
 'param==1.2.1\n',
 'pyparsing==2.0.3\n',
 'python-dateutil==2.4.0\n',
 'pytz==2014.10\n',
 'terminado==0.4\n'}
  • installed pyparsing: errors

In [8]:
fresh = []
with open("fresh.freeze") as f:
    for l in f:
        fresh.append(l)
set(magical) - set(fresh)


Out[8]:
{'holoviews==0.7\n',
 'matplotlib==1.4.3\n',
 'param==1.2.1\n',
 'pyparsing==2.0.3\n',
 'python-dateutil==2.4.0\n',
 'pytz==2014.10\n',
 'terminado==0.4\n'}
  • installed pytz: errors

In [10]:
fresh = []
with open("fresh.freeze") as f:
    for l in f:
        fresh.append(l)
set(magical) - set(fresh)


Out[10]:
{'holoviews==0.7\n',
 'matplotlib==1.4.3\n',
 'param==1.2.1\n',
 'python-dateutil==2.4.0\n',
 'terminado==0.4\n'}
  • installed terminado==0.4: errors

In [13]:
fresh = []
with open("fresh.freeze") as f:
    for l in f:
        fresh.append(l)
set(magical) - set(fresh)


Out[13]:
{'holoviews==0.7\n',
 'matplotlib==1.4.3\n',
 'param==1.2.1\n',
 'python-dateutil==2.4.0\n'}
  • installed matplotlib: errors

In [14]:
fresh = []
with open("fresh.freeze") as f:
    for l in f:
        fresh.append(l)
set(magical) - set(fresh)


Out[14]:
{'holoviews==0.7\n', 'param==1.2.1\n'}
  • installed param: errors

In [15]:
fresh = []
with open("fresh.freeze") as f:
    for l in f:
        fresh.append(l)
set(magical) - set(fresh)


Out[15]:
{'holoviews==0.7\n'}
  • installed holoviews: errors

In [18]:
fresh = []
with open("fresh.freeze") as f:
    for l in f:
        fresh.append(l)
set(magical) - set(fresh)


Out[18]:
set()

Freeze results now match exactly. Must be some other difference in the virtualenv.