Using the current docker image, we're unable to install / build TensorFlow in python3.7 with the following error:
ImportError: /usr/lib/x86_64-linux-gnu/libstdc++.so.6: version `GLIBCXX_3.4.20' not found (required by /root/miniconda2/envs/py37/lib/python3.7/site-packages/tensorflow/python/_pywrap_tensorflow_internal.so)
It seems like we need to compile using Ubuntu 16.04 libs, which will require a bump in the custom-op docker image. @yifeif would it be possible to bump the base image in the container? If so we can rebuild our release as 0.1.1 and include py37.
Alternatively, we could create a new docker image for builds.
@dynamicwebpaige Paige, tf-2.0-preview has supported python 3.7 in https://github.com/tensorflow/tensorflow/issues/25429 . Could you tell us how did you do it?
cc @gunan. Is ubuntu 14.04 being used for the build or some other docker image?
For python 3.7, we had to base our builders off of ubuntu 16.04.
Everything else uses ubuntu 14.04
We will definitely need a new image for python3.7.
Addons 0.2.0 has a py37 available on pypi
Bumping to ubuntu16+GCC5 caused issues issues as seen here in the most recent builds of TF:
https://github.com/tensorflow/tensorflow/issues/27067
This will remain unsolved until the manylinux2010 toolchain is finalized. See #119
What's the ETA for tf addons python 3.7 support?
@perfinion is there an ETA for the updated build environment / ETA when TF pip packages will be built in it? Just for reference python 3.8 release is soon.
Looks like this is almost ready:
https://github.com/tensorflow/tensorflow/commit/ec83c8a8292aa881b701928a7ffbe2305cc0e4e5
@av8ramit Could you comment on the above question?
Ideally worst case within two weeks we will transition over. Waiting on one last issue with GPU custom ops.
Most helpful comment
We will definitely need a new image for python3.7.