pip fails to install dependencies on Ubuntu 20.10

[sbrl]

If I run pip3 install -r requirements.txt on Ubuntu 20.10, I get the following error message:

ERROR: Could not find a version that satisfies the requirement tensorflow==1.9.0 (from -r requirement.txt (line 61)) (from versions: 2.2.0rc1, 2.2.0rc2, 2.2.0rc3, 2.2.0rc4, 2.2.0, 2.2.1, 2.2.2, 2.3.0rc0, 2.3.0rc1, 2.3.0rc2, 2.3.0, 2.3.1, 2.3.2, 2.4.0rc0, 2.4.0rc1, 2.4.0rc2, 2.4.0rc3, 2.4.0rc4, 2.4.0, 2.4.1, 2.5.0rc0, 2.5.0rc1, 2.5.0rc2, 2.5.0rc3, 2.5.0)
ERROR: No matching distribution found for tensorflow==1.9.0 (from -r requirement.txt (line 61))

System details:

$ uname -a
Linux DEVICE_NAME 5.8.0-50-generic #56-Ubuntu SMP Mon Apr 12 17:18:36 UTC 2021 x86_64 x86_64 x86_64 GNU/Linux
$ python --version
Python 3.8.6
$ pip --version
pip 20.1.1

[HughKu]

@sbrl Sorry for the late reply.

Python 3.8 comes as the default in Ubuntu 20.04 and as pointed out here in Tensorflow's official document: Python 3.8 support requires TensorFlow 2.2 or later.

You might wanna install python version which supports tensorflow 1.9 or port this repo to support tensorflow 2.2 up.

[sbrl]

Ah, I suspected that would be the case. I can't actually install an older version of Python very easily on Ubuntu. Hrm, I'm not sure where to start with porting this to a later version of Tensorflow for Python - my experience is mostly with Tensorflow.js (for Javascript). I've heard that the differences between Tensorflow 1.x and 2.x are quite extensive.

[pattontim]

The current dependency deltas to run with tensor 1.X and install with python 3.7. Python 3.6 didn't work because some packages were no longer served (?)

absl-py==0.4 attrs==19.2.0 gast==0.2.2 MarkupSafe==1.1.1 numpy==1.16.1 protobuf==3.6.1 tensorboard==1.13.0 tensorflow==1.13.1 Twisted==20.3.0

I won't push since 3.8+ is now the stable version. Upgrade setuptools >= 56.0.0. You may need to use a prepackaged Twisted if you get C++ build errors.