clean ffi/build with setup.py clean #569
Conversation
|
@ARF1 thanks for submitting this, I have labeled it as needing review for now. |
There was a problem hiding this comment.
I've merged main here and given this a test. It appears to do exactly as intended. For example now when running python setup.py clean we see:
$ python setup.py clean
/home/gmarkall/mambaforge/envs/numbadev/lib/python3.10/site-packages/setuptools/dist.py:493: UserWarning: Normalizing '0.40.0dev0+54.gab202ba' to '0.40.0.dev0+54.gab202ba'
warnings.warn(tmpl.format(**locals()))
running clean
removing '/home/gmarkall/numbadev/llvmlite/llvmlite.egg-info' (and everything under it)
removing '/home/gmarkall/numbadev/llvmlite/ffi/build' (and everything under it)
removing '/home/gmarkall/numbadev/llvmlite/llvmlite/__pycache__' (and everything under it)
removing '/home/gmarkall/numbadev/llvmlite/__pycache__' (and everything under it)
ffi/build is removed.
This needs a quick second review as I made a change to the super() idiom used to call super().run() but otherwise I think this is good to go.
|
CI needs #903. |
|
@gmarkall thank you for adopting this. When is the I did: So I am not yet sure when the |
|
It's created when using CMake to build. This happens on Windows, or if you build on Linux / macOS with the environment variable |
There was a problem hiding this comment.
Thanks for continuing with this @gmarkall. I think there's a hard coded path separator that needs changing, then I guess the check as described above needs to be redone (the cmake related build on windows).
Co-authored-by: stuartarchibald <[email protected]>
|
@stuartarchibald thanks for catching that - now updated. |
|
I also spotted that the clean command leaves After running the command, there is nothing left: |
|
@stuartarchibald I think this should be ready for another look now. |
There was a problem hiding this comment.
I have checked the following:
💣 zsh» git status --ignored
On branch wip/fix_setup_clean
Untracked files:
gitlog2changelog.py
test.sh
nothing added to commit but untracked files present
💣 zsh» python setup.py build
...
💣 zsh» git status --ignored
On branch wip/fix_setup_clean
Untracked files:
ffi/libllvmlite.dylib.dSYM/
gitlog2changelog.py
test.sh
Ignored files:
__pycache__/
build/
ffi/libllvmlite.dylib
ffi/libllvmlite.dylib.dSYM/Contents/Resources/
llvmlite/__pycache__/
llvmlite/binding/libllvmlite.dylib
nothing added to commit but untracked files present
💣 zsh» python setup.py clean
...
💣 zsh» git status --ignored
On branch wip/fix_setup_clean
Untracked files:
ffi/libllvmlite.dylib.dSYM/
gitlog2changelog.py
test.sh
Ignored files:
build/
nothing added to commit but untracked files present
So, only the build and ffi/libllvmlite.dylib.dSYM` remains. They contain:
💣 zsh» tree ffi/libllvmlite.dylib.dSYM/
ffi/libllvmlite.dylib.dSYM/
└── Contents
├── Info.plist
└── Resources
└── DWARF
3 directories, 1 file
💣 zsh» tree build
build
└── lib.macosx-11.1-arm64-cpython-310
└── llvmlite
├── __init__.py
├── _version.py
├── binding
│ ├── __init__.py
│ ├── analysis.py
│ ├── common.py
│ ├── context.py
│ ├── dylib.py
│ ├── executionengine.py
│ ├── ffi.py
│ ├── initfini.py
│ ├── linker.py
│ ├── module.py
│ ├── object_file.py
│ ├── options.py
│ ├── passmanagers.py
│ ├── targets.py
│ ├── transforms.py
│ └── value.py
├── ir
│ ├── __init__.py
│ ├── _utils.py
│ ├── builder.py
│ ├── context.py
│ ├── instructions.py
│ ├── module.py
│ ├── transforms.py
│ ├── types.py
│ └── values.py
├── tests
│ ├── __init__.py
│ ├── __main__.py
│ ├── customize.py
│ ├── refprune_proto.py
│ ├── test_binding.py
│ ├── test_ir.py
│ ├── test_refprune.py
│ ├── test_setup.py
│ └── test_valuerepr.py
└── utils.py
5 directories, 37 files
Would it be reasonable to assume that a clean command would remove those files also?
|
@gmarkall do you want to pickup work on this again? I think it got reasonably far and just needs a few more lines? |
|
@esc I'm not able to pick this up again in the short term. If you're aiming to tidy up the PRs and reduce the number of things open, I don't mind if you'd prefer to close / abandon this. |
|
@gmarkall it would be a shame to abandon this since it's fairly close to finish. I've placed it on the contributor self-service board, who knows, maybe we get lucky. |
Fixes #567