Recipes with versioned Python dependencies ignore requested version

[QuLogic]

We're having some trouble updating our builds to 3.5. The cartopy recipe there depends on python >=2.7,<3.5. No matter what you set CONDA_PY to, the build always attempts to use 3.5:

$ CONDA_PY=27 CONDA_NPY=19 conda build cartopy
Removing old build environment
Removing old work directory
BUILD START: cartopy-0.13.0-np19py27_2
Fetching package metadata: ........
Solving package specifications: ...............................
The following NEW packages will be INSTALLED:

    cython:          0.23.4-py35_0  
    geos:            3.4.2-3        
    libgfortran:     1.0-0          
    numpy:           1.9.3-py35_0   
    openblas:        0.2.14-3       
    openssl:         1.0.2d-0       
    owslib:          0.9.0-py35_0   
    pip:             7.1.2-py35_0   
    proj4:           4.9.1-1        
    pyepsg:          0.2.0-py35_0   
    pyshp:           1.2.1-py35_0   
    python:          3.5.0-1        
    python-dateutil: 2.4.2-py35_0   
    pytz:            2015.6-py35_0  
    readline:        6.2-2          
    requests:        2.8.1-py35_0   
    setuptools:      18.4-py35_0    
    shapely:         1.5.11-nppy35_0
    six:             1.10.0-py35_0  
    sqlite:          3.8.4.1-1      
    tk:              8.5.18-0       
    wheel:           0.26.0-py35_1  
    xz:              5.0.5-0        
    zlib:            1.2.8-0        

Even with CONDA_PY unset, the build ignores the version in the recipe and builds against 3.5:

$ CONDA_NPY=19 conda build cartopy
Removing old build environment
Removing old work directory
BUILD START: cartopy-0.13.0-np19py34_2
Fetching package metadata: ........
Solving package specifications: ...............................
The following NEW packages will be INSTALLED:

    cython:          0.23.4-py35_0  
    geos:            3.4.2-3        
    libgfortran:     1.0-0          
    numpy:           1.9.3-py35_0   
    openblas:        0.2.14-3       
    openssl:         1.0.2d-0       
    owslib:          0.9.0-py35_0   
    pip:             7.1.2-py35_0   
    proj4:           4.9.1-1        
    pyepsg:          0.2.0-py35_0   
    pyshp:           1.2.1-py35_0   
    python:          3.5.0-1        
    python-dateutil: 2.4.2-py35_0   
    pytz:            2015.6-py35_0  
    readline:        6.2-2          
    requests:        2.8.1-py35_0   
    setuptools:      18.4-py35_0    
    shapely:         1.5.11-nppy35_0
    six:             1.10.0-py35_0  
    sqlite:          3.8.4.1-1      
    tk:              8.5.18-0       
    wheel:           0.26.0-py35_1  
    xz:              5.0.5-0        
    zlib:            1.2.8-0        

[QuLogic]

Also, if you change the dependency to python >=2.7, you still get this behaviour. Only if you remove the version entirely does the build use the correct dependencies for building.

[pelson]

Note, this relates to a discussion ongoing in #650.

[QuLogic]

The Python dependency doesn't even use x.x though, unless it's because of its implementation that issues occur?

[mwcraig]

@QuLogic -- try this:

- python # or python x.x, either works
- python >=2.7,<3.5

I think it has always been the case that conda has required python by itself to use CONDA_PY. You will need to use the same specification in both run and build for this to work, I think.

[QuLogic]

That would be not too intuitive, but unfortunately it does not work. It's okay for 2.7 and 3.4, but if you try CONDA_PY=35, it still builds even though that should be an error.

[github-actions]

Hi there, thank you for your contribution!

This issue has been automatically marked as stale because it has not had recent activity. It will be closed automatically if no further activity occurs.

If you would like this issue to remain open please:

1. Verify that you can still reproduce the issue at hand
2. Comment that the issue is still reproducible and include:
   - What OS and version you reproduced the issue on
   - What steps you followed to reproduce the issue

NOTE: If this issue was closed prematurely, please leave a comment.

Thanks!