New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
tensorflow: 1.15.2 -> 1.15.3 #93177
tensorflow: 1.15.2 -> 1.15.3 #93177
Conversation
- with pytest.raises(TypeError): | ||
+ with pytest.raises(ValueError): |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I don't feel like this is a nixpkgs-specific issue, is there a way to upstream it?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I have created an issue upstream: dgasmith/opt_einsum#147
Tensorflow doesn't seem to build for me with python 3.7 or 3.6, did it build for you when you opened this PR? And einsum was bumped to 3.3.0 in #93907 |
seems to fail with:
With how often tensorflow_1 breaks. It's largely amazing to me that it ever builds. |
I'm skeptical you can get tensorflow 1.15 to work with anything but bazel 0.26 - see https://github.com/tensorflow/tensorflow/blob/v1.15.3/configure.py#L52-L53. Is tensorflow 1 still being actively used in the community? I was hoping we could fix tensorflow 2 derivations, then remove the tensorflow 1 stuff. |
looks like the older bazel packages were restored, it should be able to call bazel with the older versions. |
We are still using TF 1.x. Please keep it around, I hope to find a revision that works. |
There are still projects that haven't migrated, e.g. photoprism/photoprism#222 |
I started this PR before leaving for vacation and I just got back. I can't really remember, but I think it did not finish successfully.
I am not trying to build tensorflow with an updated bazel version, it's just that dm-sonnet is build with bazel 3 and tensorflow needs to be build as a dependency. I don't fully understand why. 🤔 |
Apparently, this is caused by an ABI change of numpy 1.19.0 - see tensorflow/tensorflow#40688 |
This PR should be unblocked by #98083 |
56a6e78
to
5f302ca
Compare
5f302ca
to
5a20b19
Compare
Since #98083 has been merged, this is ready to review. |
Tensorflow is already at 1.15.4 |
Motivation for this change
tensorflow 1.15.3 was released.
The first commit fixes the build for opt-einsum (see PR #93107), but I also wanted to check whether tensorflow builds.Actually, my main motivation for this is that I want to upgrade bazel to 3.4.1, but they removed the
--incompatible_disable_deprecated_attr_params
command line flag, which is used by dm-sonnet. So I tried to upgrade dm-sonnet, hoping that they would not need this flag anymore. But that did not work because it depends on tensorflow and tensorflow did not build because of the broken test in opt-einsum.Things done
sandbox
innix.conf
on non-NixOS linux)nix-shell -p nixpkgs-review --run "nixpkgs-review wip"
./result/bin/
)nix path-info -S
before and after)