Kyle Beauchamp

I've also found that this is a problem when building the bundled tar dependencies such as HTSLib, which may have their own hard-coded gcc calls in their Makefiles. See, e.g.,...

As a starting point, I made a WIP PR to get a conda package for TF2 nightly: https://github.com/conda-forge/tensorflow-feedstock/pull/68

1. I'm 90% sure the answer is yes, but there will be issues with GPU support given the state of nvidia licensing (doesn't support OSS projects like conda-forge). 2. We're...

I also think something like this might be useful for keeping track of pandas metadata for future use.

FWIW, it's easy enough to do this in some settings: ``` X = patsy.dmatrices("y ~ col0 * col1 - 1", data, return_type="dataframe")[1] X_csr = scipy.sparse.csr_matrix(X.values) ```

Agreed. My particular use case was for doing MCMC with the sparse matrix via pymc2, however, so I was able to get massive speedups when taking products of model parameters...

LGTM, do we want to copy in the MRE from #317 as a test?

I'm not super aware of possible side effects of this change though, so someone else probably needs approve.

``` pytables 3.2.2 np110py27_0 pymc 2.3.6 np110py27_0 ```

Here you go. AFAIK, the key ingredient is something about the observed variable or the deterministic. Without those, things work fine in HDF. ``` import pymc as pm n_vars =...