Tensorflow 2.13.0 cannot be imported after install via poetry due to missing wheel metadata

[dre-hh]

Issue type

Bug

Have you reproduced the bug with TensorFlow Nightly?

No

Source

binary

TensorFlow version

2.13.0

Custom code

No

OS platform and distribution

macOS-13.3-arm64-arm-64bit

Mobile device

No response

Python version

3.11

Bazel version

No response

GCC/compiler version

No response

CUDA/cuDNN version

No response

GPU model and memory

No response

Current behavior?

After installing tensorflow 2.13.0 with poetry it cannot be imported

Standalone code to reproduce the issue

See python-poetry/poetry#8271

According to @dimbleby, the issue happens due to missing wheel metadata.

please encourage the tensorflow folk to publish consistent metadata in all of their wheels, with platform-specific variations described by markers

Tensorflow folks, please publish consistent metadata ;)

poetry add tensorflow==2.13.0
poetry shell
python3 
import tensorflow

Relevant log output

ModuleNotFoundError: No module named 'tensorflow

[Varsha-anjanappa]

Hi @dre-hh ,

As you can see this issue. Yes tensorflow's metadata does not express tensorflow-macos dependency. As a workaround you can explicitly add tensorflow-macos which solves the issue -
poetry add tensorflow-macos==2.13.0

Thank you!!

[win845]

Thx for reply. Unfortunately this is not a good workaround. People nowadays develop on arm macs and deploy on x86. That means , one cannot produce a lockfile for the whole team to install same package versions cross platform . It is a problem which is hard to solve in python but poetry addresses quite well.

There is also another problem:

If you check the Metadata inside the pip site_packages. On arm linux tensorflow
also installs a different package ( tensorflow aws). This dependency is also not expressed on pypa json Metadata . I suppose poetry is only using that, and will install wrong packages on linux arm as well.

Could you please publish consistent Metadata to pypa in same way it is expressed in Metadata file when the package is downloaded

[github-actions]

This issue is stale because it has been open for 7 days with no activity. It will be closed if no further activity occurs. Thank you.

[dre-hh]

@Varsha-anjanappa any update on the idea adding same dependencies to pypi metadata as in the pip package?
Is this maybe tracked on another issue?
Is there a reason which prevents tensorflow team from doing this?
Without this metadata , the user of the package has to figure out himself which platform specific packages to install

[sachinprasadhs]

@kulinseth , Is it something to do with Tensorflow MacOS release?

[dre-hh]

@sachinprasadhs, no it is the tensorflow package itself.

cd  python-3.11/lib/python3.11/site-packages/tensorflow-2.13.0.dist-info
cat METADATA

=>
Description-Content-Type: text/markdown
Requires-Dist: tensorflow-macos (==2.13.0) ; platform_system == "Darwin" and platform_machine == "arm64"
Requires-Dist: tensorflow-cpu-aws (==2.13.0) ; platform_system == "Linux" and (platform_machine == "arm64" or platform_machine == "aarch64")
Requires-Dist: tensorflow-intel (==2.13.0) ; platform_system == "Windows"

This metadata was not published to pypi

$ curl -s https://pypi.org/pypi/tensorflow/2.13.0/json | jq '.info.requires_dist'
[
  "absl-py (>=1.0.0)",
  "astunparse (>=1.6.0)",
  "flatbuffers (>=23.1.21)",
  "gast (<=0.4.0,>=0.2.1)",
  "google-pasta (>=0.1.1)",
  "h5py (>=2.9.0)",
  "libclang (>=13.0.0)",
  "numpy (<=1.24.3,>=1.22)",
  "opt-einsum (>=2.3.2)",
  "packaging",
  "protobuf (!=4.21.0,!=4.21.1,!=4.21.2,!=4.21.3,!=4.21.4,!=4.21.5,<5.0.0dev,>=3.20.3)",
  "setuptools",
  "six (>=1.12.0)",
  "termcolor (>=1.1.0)",
  "typing-extensions (<4.6.0,>=3.6.6)",
  "wrapt (>=1.11.0)",
  "grpcio (<2.0,>=1.24.3)",
  "tensorboard (<2.14,>=2.13)",
  "tensorflow-estimator (<2.14,>=2.13.0)",
  "keras (<2.14,>=2.13.1)",
  "tensorflow-io-gcs-filesystem (>=0.23.1) ; platform_machine != \"arm64\" or platform_system != \"Darwin\""
]

Please republish a new version of the package with consitent medata as pip downloads into tensorflow-2.13.0.dist-info/METADATA file