Failed building wheel for dm-tree

[haydenshively]

Installing garage dependencies into a fresh conda environment (Python 3.8) via pip install --no-cache -e ."[mujoco,dm_control]" yields the following errors related to dm-tree. These may or may not be specific to my computer/macOS, but they are reproducible in every fresh conda env that I make. If instead I try to install from PyPI (pip install --no-cache -e garage"[mujoco,dm_control]"), it says Packages installed from PyPI cannot depend on packages which are not also hosted on PyPI since dm_control is from deepmind's repo.

ERROR: Command errored out with exit status 1:
command: /Users/haydenshively/anaconda3/envs/Garage/bin/python -u -c 'import sys, setuptools, tokenize; sys.argv[0] = '"'"'/private/var/folders/hz/1l2dtvyj4dv35mys1v9y2nd40000gn/T/pip-install-mk32dmoe/dm-tree/setup.py'"'"'; file='"'"'/private/var/folders/hz/1l2dtvyj4dv35mys1v9y2nd40000gn/T/pip-install-mk32dmoe/dm-tree/setup.py'"'"';f=getattr(tokenize, '"'"'open'"'"', open)(file);code=f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, file, '"'"'exec'"'"'))' bdist_wheel -d /private/var/folders/hz/1l2dtvyj4dv35mys1v9y2nd40000gn/T/pip-wheel-5hwif72_
cwd: /private/var/folders/hz/1l2dtvyj4dv35mys1v9y2nd40000gn/T/pip-install-mk32dmoe/dm-tree/
Complete output (13 lines):
running bdist_wheel
running build
running build_py
creating build
creating build/lib.macosx-10.9-x86_64-3.8
creating build/lib.macosx-10.9-x86_64-3.8/tree
copying tree/init.py -> build/lib.macosx-10.9-x86_64-3.8/tree
copying tree/tree_test.py -> build/lib.macosx-10.9-x86_64-3.8/tree
copying tree/tree_benchmark.py -> build/lib.macosx-10.9-x86_64-3.8/tree
running build_ext
bazel build //tree:_tree --symlink_prefix=build/temp.macosx-10.9-x86_64-3.8/bazel- --compilation_mode=opt
unable to execute 'bazel': No such file or directory
error: command 'bazel' failed with exit status 1

---

ERROR: Command errored out with exit status 1:
command: /Users/haydenshively/anaconda3/envs/Garage/bin/python -u -c 'import sys, setuptools, tokenize; sys.argv[0] = '"'"'/private/var/folders/hz/1l2dtvyj4dv35mys1v9y2nd40000gn/T/pip-install-mk32dmoe/dm-tree/setup.py'"'"'; file='"'"'/private/var/folders/hz/1l2dtvyj4dv35mys1v9y2nd40000gn/T/pip-install-mk32dmoe/dm-tree/setup.py'"'"';f=getattr(tokenize, '"'"'open'"'"', open)(file);code=f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, file, '"'"'exec'"'"'))' install --record /private/var/folders/hz/1l2dtvyj4dv35mys1v9y2nd40000gn/T/pip-record-bgk0rzmx/install-record.txt --single-version-externally-managed --compile --install-headers /Users/haydenshively/anaconda3/envs/Garage/include/python3.8/dm-tree
cwd: /private/var/folders/hz/1l2dtvyj4dv35mys1v9y2nd40000gn/T/pip-install-mk32dmoe/dm-tree/
Complete output (13 lines):
running install
running build
running build_py
creating build
creating build/lib.macosx-10.9-x86_64-3.8
creating build/lib.macosx-10.9-x86_64-3.8/tree
copying tree/init.py -> build/lib.macosx-10.9-x86_64-3.8/tree
copying tree/tree_test.py -> build/lib.macosx-10.9-x86_64-3.8/tree
copying tree/tree_benchmark.py -> build/lib.macosx-10.9-x86_64-3.8/tree
running build_ext
bazel build //tree:_tree --symlink_prefix=build/temp.macosx-10.9-x86_64-3.8/bazel- --compilation_mode=opt
unable to execute 'bazel': No such file or directory
error: command 'bazel' failed with exit status 1

[haydenshively]

It seems that dm-tree is not compatible with Python 3.8: google-deepmind/dm_control#135
We should probably specify compatible Python versions in the installation instructions.

[ryanjulian]

Great catch! The best way for us to specify this is to update setup.py

In particular, you'll want python_requires='>=3.5,<3.8

why don't you give it a shot?

[ryanjulian]

(you should also leave a comment explaining 3.8 is unsupported, with a link to the issue you found)

[gitanshu]

The issue is that compiling dm-tree from source requires installing bazel. So it's not that we are not compatible with python 3.8, but that on 3.8, we have an extra system dependency. Do you think we should still restrict our supported range of python versions for that?

[haydenshively]

I’m happy to modify the setup.py either way. It’s be great to support 3.8 by adding the bazel dependency, but on the other hand we already have a lot of dependencies, and the bazel binary is rather large (~120mb). Let me know which way to go.

[ryanjulian]

The root of the problem is not bazel, but DeepMind failing to upload a pre-compiled wheel for Python 3.8 to PyPI, causing pip to revert to trying to compile some native extensions from source code (which requires bazel).

We should probably just wait for them to upload the wheel.

Until then, we should at least properly declare that not all configurations support Python 3.8.

[ryanjulian]

This should be fixed -- check again?

[haydenshively]

Yep. Looks like it's working now

[xwy0303]

Hello. I have the same problem with you, but The version of python on my computer is 3.6. Do you know why? Thanks!

[mi-heinsch]

Hey, I have the same issue with Python 3.6.9.. Anyone has a fix already?

[aymen5507]

any update?

[shikharmittal04]

Same issue here. I have Python 3.6.8

[rauki34]

oh,damn. same problem in 3.12 when I use pip install keras. Where is the solution?

[shikharmittal04]

Upgrading to Python 3.11.5 solved the problem for me.

[rauki34]

Upgrading to Python 3.11.5 solved the problem for me.

success right now amazingly while I still do not know what problem it happened.