Closed sdumpling closed 3 years ago
Hi @sdumpling we will eventually be adding python3.9 support but it is not a high priority for us. However if there are more requests for 3.9 support we can re-prioritize this.
We are blocked on TF supporting 3.9. If they build for py3.9 in TF 2.5, we will be able to add appropriate tests.
With today's release of TF 2.5.0 there is TensorFlow support for Python 3.9. For the moment this seems to not be compatible with the stable release of reverb
. (I have raised an issue about this #42)
The most recent Reverb release is compatible with TF 2.5.0 🎉
We do not have a 3.9 release yet. I am looking at adding that build this week now that TF 2.5.0 supports it. Hopefully it is straight forward...although rarely is anything straight forward. Assigning this bug to me.
Quick update @IanRDavies Python 3.9 looks good. All test passed and the docker upgrade was super easy for once. I am making a couple small PRs/CLs. 99% chance I push the 3.9 build this week as 0.3.1 along with fresh 3.6 through 3.8.
@IanRDavies Would you mind confirming that you are now able to use Reverb with Python 3.9?
@acassirer I can confirm that reverb runs with Python 3.9. I haven't exhaustively tested all features but for all of my use cases it works!
Awesome! Thanks @IanRDavies for verifying!
Installing reverb for python3.9 failed ERROR: Could not find a version that satisfies the requirement dm-reverb>=0.2.0; extra == "reverb" (from tf-agents[reverb]) (from versions: none) ERROR: No matching distribution found for dm-reverb>=0.2.0; extra == "reverb" (from tf-agents[reverb])
I also noticed that this package is marked for python3.{6,7,8} only. Is there any plan for supporting python3.9?
Thanks.