Mihai Maruseac
Mihai Maruseac
This is still an issue today.
https://github.com/tensorflow/tensorflow/issues/57671
I was able to build TF inside Docker via `yes "" | python configure.py`
The `configure.py` script is extremely convoluted and grew organically. If someone were to provide a PR documenting the features I think it can be accepted.
Note that since 2 years ago we no longer use `./configure` / `configure.py` for the standard builds. These options are now in `.bazelrc` by default and you should only use...
TF 2.6 is no longer in the window of supported versions
Main branch: https://github.com/tensorflow/tensorflow/blob/92d57e61d41f8e6fca50d582ac7a56046a391591/tensorflow/workspace2.bzl#L493-L500 r2.10: https://github.com/tensorflow/tensorflow/blob/359c3cdfc5fabac82b3c70b3b6de2b0a8c16874f/tensorflow/workspace2.bzl#L491-L498 r2.9: https://github.com/tensorflow/tensorflow/blob/18960c44ad3f5219c22dca55f842912dbce78a07/tensorflow/workspace2.bzl#L483-L490 r2.8: https://github.com/tensorflow/tensorflow/blob/92a51d52ad199319e4f9de83fcbe970151dfed7e/tensorflow/workspace2.bzl#L490-L497 So, if you take the last release on each of these branches you are not affected
Then you might be affected on all supported versions. TF does not generate VEX statements at this point, so we cannot be certain.
TF 2.6 is no longer in the windows of supported versions.