Cannot import GPy on mybinder with Python 3.5

[glouppe]

Hi,

I am trying to make public on mybinder.org some notebooks of mine where GPy is used. Installation works fine, but importing GPy fails with the following error:

/home/main/anaconda2/envs/binder/lib/python3.5/site-packages/GPy/core/__init__.py in <module>()                                    
      6 from . import parameterization                                                                                             
      7                                                                                                                            
----> 8 from .gp import GP                                                                                                         
      9 from .svgp import SVGP                                                                                                     
     10 from .sparse_gp import SparseGP                                                                                            

/home/main/anaconda2/envs/binder/lib/python3.5/site-packages/GPy/core/gp.py in <module>()                                          
      3                                                                                                                            
      4 import numpy as np                                                                                                         
----> 5 from .. import kern                                                                                                        
      6 from GPy.core.model import Model                                                                                           
      7 from paramz import ObsAr                                                                                                   

/home/main/anaconda2/envs/binder/lib/python3.5/site-packages/GPy/kern/__init__.py in <module>()                                    
      6    :private-members:                                                                                                       
      7 """                                                                                                                        
----> 8 from .src.kern import Kern                                                                                                 
      9 from .src.add import Add                                                                                                   
     10 from .src.prod import Prod                                                                                                 

/home/main/anaconda2/envs/binder/lib/python3.5/site-packages/GPy/kern/src/__init__.py in <module>()                                
----> 1 from . import psi_comp                                                                                                     

/home/main/anaconda2/envs/binder/lib/python3.5/site-packages/GPy/kern/src/psi_comp/__init__.py in <module>()                       
     66                                                                                                                            
     67 from . import ssrbf_psi_gpucomp                                                                                            
---> 68 from .rbf_psi_gpucomp import PSICOMP_RBF_GPU                                                                               

/home/main/anaconda2/envs/binder/lib/python3.5/site-packages/GPy/kern/src/psi_comp/rbf_psi_gpucomp.py in <module>()                
      6 from paramz.caching import Cache_this                                                                                      
      7 from . import PSICOMP_RBF                                                                                                  
----> 8 from ....util import gpu_init                                                                                              
      9                                                                                                                            
     10 gpu_code = """                                                                                                             

/home/main/anaconda2/envs/binder/lib/python3.5/site-packages/GPy/util/__init__.py in <module>()                                    
      7 from . import squashers                                                                                                    
      8 from . import warping_functions                                                                                            
----> 9 from . import datasets                                                                                                     
     10 from . import mocap                                                                                                        
     11 from . import decorators                                                                                                   

/home/main/anaconda2/envs/binder/lib/python3.5/site-packages/GPy/util/datasets.py in <module>()                                    
     55 if not (on_rtd):                                                                                                           
     56     path = os.path.join(os.path.dirname(__file__), 'data_resources.json')                                                  
---> 57     json_data=open(path).read()                                                                                            
     58     data_resources = json.loads(json_data)                                                                                 
     59                                                                                                                            

/home/main/anaconda2/envs/binder/lib/python3.5/encodings/ascii.py in decode(self, input, final)                                    
     24 class IncrementalDecoder(codecs.IncrementalDecoder):                                                                       
     25     def decode(self, input, final=False):                                                                                  
---> 26         return codecs.ascii_decode(input, self.errors)[0]                                                                  
     27                                                                                                                            
     28 class StreamWriter(Codec,codecs.StreamWriter):                                                                             

UnicodeDecodeError: 'ascii' codec can't decode byte 0xc3 in position 7276: ordinal not in range(128)                   

GPy is installed from the devel branch within the following conda environment https://github.com/diana-hep/carl-notebooks/blob/master/environment.yml#L14

The tricky issue is that I cant reproduce the problem locally. The same environment file works fine on my machine :(

CC'ing some Binder people to the rescue @rgbkrk @freeman-lab

[glouppe]

I am closing this. Sorry for the noise. It seems it is rather related to binder. I have the exact same error with a different independent package.

[lawrennd]

Not pure noise, I just found out about mybinder.org :-)

Neil

On Fri, Mar 4, 2016 at 4:43 PM, Gilles Louppe notifications@github.com wrote:

I am closing this. Sorry for the noise. It seems it is rather related to binder. I have the exact same error with a different independent package.

— Reply to this email directly or view it on GitHub https://github.com/SheffieldML/GPy/issues/319#issuecomment-192350683.

[mikecroucher]

mybinder is cool. Found this a couple of weeks ago http://ivory.idyll.org/blog/2016-mybinder.html

[ken-okabe]

Hi, I have the same issue on console and JupyterNotebook with Python 3.5. Having tried everything I found on the web, but for some reason,

detasets.py

 json_data=open(path).read()

always calls .../lib/python3.5/encodings/ascii.py

so, all I could do is to modify the sourcecode to give encoding option enforcing to use 'utf-8' of the function in detasets.py as below:

if not (on_rtd):
    path = os.path.join(os.path.dirname(__file__), 'data_resources.json')
    json_data=open(path, encoding='utf-8').read()
    data_resources = json.loads(json_data)

if not (on_rtd):
    path = os.path.join(os.path.dirname(__file__), 'football_teams.json')
    json_data=open(path, encoding='utf-8').read()
    football_dict = json.loads(json_data)

It would be pretty that the developer give the encoding='utf-8' option from next update, otherwise I probably have to make this patch on every release :(

Thanks!

[lawrennd]

Hi Ken,

We can easily make that change, but it's much better if you submit it by a pull request so you get the correct credit for fixing the issue and we keep the paper trail for what the problem was.

https://help.github.com/articles/using-pull-requests/

To understand correctly, is it an issue with reading those files on Python3.5 in general?

Neil

On Fri, Apr 22, 2016 at 2:15 AM, Ken OKABE notifications@github.com wrote:

Hi, I have the same issue on console and JupyterNotebook with Python 3.5. Having tried everything I found on the web, but for some reason,

detasets.py

json_data=open(path).read()

always calls .../lib/python3.5/encodings/ascii.py

so, all I could do is to modify the sourcecode to give encoding option of the function in detasets.py as below:

if not (on_rtd): path = os.path.join(os.path.dirname(file), 'data_resources.json') json_data=open(path, encoding='utf-8').read() data_resources = json.loads(json_data)

if not (on_rtd): path = os.path.join(os.path.dirname(file), 'football_teams.json') json_data=open(path, encoding='utf-8').read() football_dict = json.loads(json_data)

It would be pretty that the developer give the encoding='utf-8' option from next update, otherwise I probably have to make this patch on every release :(

Thanks!

— You are receiving this because you commented. Reply to this email directly or view it on GitHub https://github.com/SheffieldML/GPy/issues/319#issuecomment-213277974

[ken-okabe]

Hi Neil,

Thank you for this stunning project.

I made a pull request: https://github.com/SheffieldML/GPy/pull/378

To understand correctly, is it an issue with reading those files on Python3.5 in general?

Basically, I'm a JavaScript/node developer with insufficient experience for Python, so I'm sorry I could not figure out how this error occurred to my environment.

So far, I have installed many other packages via pip, including tensor-flow and theano, where they could download and read the data-files without problems. For very unknown reason, unfortunately GPy is the only one that has this issue so far, and I also suspected my own Python3.5 global encoding configuration, and searched the web and tripple-checked, but seems no problem.

Anyway, encoding issue seems notorious in Python environment, and this fix should be useful in general I believe.

Thanks! Ken

[lawrennd]

Thanks so much Ken! It's not really me, it's a community effort :-)

Max (@mzwiessele) is really the one driving things forward with the pull requests and taking the lead on the releases.

Encoding does indeed seem to be a nightmare with python, and the problem is compounded by the fact that we are currently releasing code for python 2 & 3.

On Fri, Apr 22, 2016 at 6:39 AM, Ken OKABE notifications@github.com wrote:

Hi Neil,

Thank you for this stunning project.

I made a pull request: #378 https://github.com/SheffieldML/GPy/pull/378

To understand correctly, is it an issue with reading those files on Python3.5 in general?

Basically, I'm a JavaScript/node developer with insufficient experience for Python, so I'm sorry I could not figure out how this error occurred to my environment.

So far, I have installed many other packages via pip, including tensor-flow and theano, where they could download and read the data-files without problems. For very unknown reason, unfortunately GPy is the only one that has this issue so far, and I also suspected my own Python3.5 global encoding configuration, and searched the web and tripple-checked, but seems no problem.

Anyway, encoding issue seems notorious in Python environment, and this fix should be useful in general I believe.

Thanks! Ken

— You are receiving this because you commented. Reply to this email directly or view it on GitHub https://github.com/SheffieldML/GPy/issues/319#issuecomment-213373932

[ken-okabe]

Hi Neil,

It's also my pleasure to have been able to contribute this project even a little :)

Actually, while studying machine learning on the basis of not NN but probabilistic in a generalized framework, I have read your papers on Gaussian process latent variable models. and felt truly epoch-making research. In fact, I'm looking forward to your book about GP for next generation as the successor of PRML by Christopher M. Bishop!