sync

[ocefpaf]

This recipe is quite different from the one here and maybe we should put this in a branch and slowly transition to it? (Or maybe just rip the band-aid and get a new and shiny recipe :wink:)

@jjhelmus looks like you are the one that wrote this version, do you mind taking a look?

[conda-forge-linter]

Hi! This is the friendly automated conda-forge-linting service.

I wanted to let you know that I linted all conda-recipes in your PR (recipe) and found some lint.

Here's what I've got...

For recipe:

• The home item is expected in the about section.
• The license item is expected in the about section.
• The summary item is expected in the about section.

For recipe:

• It looks like the 'keras' output doesn't have any tests.
• It looks like the 'keras-gpu' output doesn't have any tests.

[ocefpaf]

@conda-forge-admin, please rerender

[conda-forge-linter]

Hi! This is the friendly automated conda-forge-webservice.

I tried to re-render for you, but it looks like there was nothing to do.

[ocefpaf]

I guess we'll need to wait on the new compilers :-(

ImportError: /lib64/libc.so.6: version `GLIBC_2.17' not found (required by /home/conda/feedstock_root/build_artifacts/keras_split_1534444619053/_test_env_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placehold_placeh/lib/python3.6/site-packages/tensorflow/python/_pywrap_tensorflow_internal.so)

[mariusvniekerk]

So we might not have to wait on the new compilers, but we can't import tensorflow in the very old centos image we use.

[ocefpaf]

So we might not have to wait on the new compilers, but we can't import tensorflow in the very old centos image we use.

I realized that after I send the comment. @jjhelmus and @msarahan told me that already a few times and I always forget.

[jjhelmus]

I bit of background on what this recipe is doing. This produced three Keras related packages: keras-base contains the actual Python code and does not require a backend, keras adds a requirement on keras and keras-gpu add a requirements on tensorflow-gpu. The idea here is that most users will want to install keras and have a working environment with the Tensorflow backend. If you have a GPU you can install keras-gpu. More advanced users can install keras-base and then install a backend on their own, perhaps PyTorch.

I think the import tests will need to be skipped on linux and os-x. The tensorflow conda packages repackaging the wheels from PyPI which require newer versions of on Linux and macOS than are used on the CI services.

[jjhelmus]

Should the activation scripts be removed as well? They are not used in the recipe.

[conda-forge-linter]

Hi! This is the friendly automated conda-forge-linting service.

I wanted to let you know that I linted all conda-recipes in your PR (recipe) and found some lint.

Here's what I've got...

For recipe:

• The home item is expected in the about section.
• The license item is expected in the about section.
• The summary item is expected in the about section.

For recipe:

• It looks like the 'keras' output doesn't have any tests.
• It looks like the 'keras-gpu' output doesn't have any tests.

[ocefpaf]

Should the activation scripts be removed as well? They are not used in the recipe.

Missed those! Just updated and rebased.