Created wrf-python recipe. #2229
Conversation
|
Hi! This is the friendly automated conda-forge-linting service. I just wanted to let you know that I linted all conda-recipes in your PR ( |
|
Is there anything else I need to do to convert this to a feedstock? Is this all automated from here? I have a talk at AMS on Monday, and was hoping to get a conda-forge package available for people that want to try it out. This package is related to NCL, PyNIO, and PyNGL. Thanks! |
Merge and wait 😄
Yep.
I am merging this now and hopefully the CIs will process this in time. I have a few minor comments that we can address later in the feedstock. Good luck with your talk! |
|
|
||
| about: | ||
| home: https://github.com/NCAR/wrf-python | ||
| license: "NCL Source Code" |
There was a problem hiding this comment.
Please add a license_file here. They are specially important for custom license like this one.
There was a problem hiding this comment.
No problem. I'll add the license file to the recipe directory and follow the instructions in the docs.
| - gcc # [unix] | ||
| - libgcc # [unix] | ||
| - xarray | ||
| - vc 9 # [win and py27] |
There was a problem hiding this comment.
You don't need vcs. Because this is depends on python at both build and runthe right vc will be always activated.
There was a problem hiding this comment.
I think I added this when I was trying to build against Python 3.5 on Windows and not having much luck. I'll remove it.
| - wrapt | ||
| - m2w64-toolchain # [win] | ||
| - gcc # [unix] | ||
| - libgcc # [unix] |
There was a problem hiding this comment.
If you need Fortran try the libgfortran instead of libgcc. (On Linux you may still need the libgcc though if you need libquadmath.)
There was a problem hiding this comment.
I wasn't sure about this one. I think libgfortran is for OSX only, and libgcc is supposed to have the fortran symbols in them, and then one day that stopped working, so I just pulled the whole gcc in as well. I'll play around with it some more.
There was a problem hiding this comment.
This is quite a mess in conda-forge, but here is the gist of it:
- if you need Fortran and
libquadmathyou needlibgccon Linux andlibgfortranon OS X - if you only need Fortran you are OK with
libgfortranon both - if you need
libgompyou needlibgccon both, but then your package may be inconsistent on OS X because there people are usingclanginstead ofgcc.
| - wrapt | ||
| - m2w64-toolchain # [win] | ||
| - gcc # [unix] | ||
| - libgcc # [unix] |
There was a problem hiding this comment.
You don't need libgcc here because gcc from above installs it for you. libgcc is a run time dependency only.
|
|
||
| requirements: | ||
| build: | ||
| - pip |
There was a problem hiding this comment.
You probably need setuptools, no?
There was a problem hiding this comment.
Correct. I originally used pip, but during the try/fail cycle, I changed to setuptools. Will update the feedstock.
| - pip | ||
| - numpy x.x | ||
| - wrapt | ||
| - m2w64-toolchain # [win] |
There was a problem hiding this comment.
Not sure if you need everything from m2w64-toolchain to get Fortran on Windows. @mingwandroid probably know better.
There was a problem hiding this comment.
Google searching led me to that solution.
Also, I can't build with Python 3.5. I think the mingw compiler now works with Python 3.5, but numpy distutils (and python distutils) needs patching. It looks like this is being addressed in numpy 1.13.
| build: | ||
| number: 1 | ||
| detect_binary_files_with_prefix: true | ||
| skip: True # [win32 or np>=112 or (win and py>=35)] |
There was a problem hiding this comment.
I usually have success building Fortran extensions on win32. Just curious why you skipped it.
There was a problem hiding this comment.
Win32 didn't work for some reason. I'm not sure if it's the m2w64-toolchain, numpy's distutils, or me supplying the wrong compiler argument. It failed with a lot of missing python symbols, which should have been in libpython, so I don't think it got linked in correctly. Since I'm a little pressed for time, I decided to skip it, and will try to figure it out after AMS.
There was a problem hiding this comment.
Not sure why ATM, but something about this selector is causing problems in feedstock conversion. May require some investigation.
xref: https://travis-ci.org/conda-forge/staged-recipes/builds/193381312#L319-L328
No description provided.