Closed jbransen closed 1 year ago
Do you feel like updating the pinned version of TF at the same time? Then you don't have to worry about backward compatibility. Some of the recent commits are such updates, so you have a template to start with.
Yes, that makes sense, I believe I did that now (though I did not actually run the Docker containers, but I did verify that the base images exist so I have good hopes that this is right).
Thanks!
Commit https://github.com/tensorflow/tensorflow/commit/78d8fdde71b9a0dc2e1919194f071ba8c222e1d5 renamed the status fields from the
TF_
prefix toTSL_
, though there are define's in place to keep the old names, the Haskell FFI generator picks up the new names, leading to compile errors.To fix compilation, and be forward compatible, I renamed the fields, which means this branch is not compatible with older tensorflow versions, but I don't know how to make it so without ugly hacks.