8000 Rebuild for pypy37 by regro-cf-autotick-bot · Pull Request #66 · conda-forge/pytorch-cpu-feedstock · GitHub
[go: up one dir, main page]

Skip to content

Rebuild for pypy37 #66

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Conversation

regro-cf-autotick-bot
Copy link
Contributor

This PR has been triggered in an effort to update pypy37.

Notes and instructions for merging this PR:

  1. Please merge the PR only after the tests have passed.
  2. Feel free to push to the bot's branch to update this PR if needed.

Please note that if you close this PR we presume that the feedstock has been rebuilt, so if you are going to perform the rebuild yourself don't close this PR until the your rebuild has been merged.

This package has the following downstream children:

And potentially more.

If this PR was opened in error or needs to be updated please add the bot-rerun label to this PR. The bot will close this PR and schedule another one. If you do not have permissions to add this label, you can use the phrase @conda-forge-admin, please rerun bot in a PR comment to have the conda-forge-admin add it for you.

This PR was created by the regro-cf-autotick-bot.
The regro-cf-autotick-bot is a service to automatically track the dependency graph, migrate packages, and propose package version updates for conda-forge. If you would like a local version of this bot, you might consider using rever. Rever is a tool for automating software releases and forms the backbone of the bot's conda-forge PRing capability. Rever is both conda (conda install -c conda-forge rever) and pip (pip install re-ver) installable.
Finally, feel free to drop us a line if there are any issues!
This PR was generated by https://github.com/regro/autotick-bot/actions/runs/1139449109, please use this URL for debugging

@conda-forge-linter
Copy link

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe) and found it was in an excellent condition.

@hmaarrfk
Copy link
Contributor

@mattip any clue on how to make this compatible?

@mattip
Copy link
mattip commented Aug 20, 2021

Here is the error:

.../torch/csrc/jit/python/python_tracer.cpp: In function 'std::vector<torch::jit::StackEntry> torch::jit::tracer::_pythonCallstack()':
.../torch/csrc/jit/python/python_tracer.cpp:39:20: error: 'PyFrameObject {aka struct _frame}' has no member named 'f_back'
frame = frame->f_back;

PyPy does not expose a f_back field in its PyFrameObject. The goal of that function is to get a complete stack. I think it would be OK to only expose the current stack level by adding a patch like

#if defined(PYPY_VERSION_NUM)
frame == nullptr;
#else
frame = frame->f_back;
#endif

I am checking how hard it would be to add f_back. If I can get it going by the next release then the first line would become

#if defined(PYPY_VERSION_NUM) && PYPY_VERSION_NUM < 0x07030600

@hmaarrfk
Copy link
Contributor

For what its worth, I'm adding:

#if !defined(PYPY_VERSION_NUM)
    size_t line = PyCode_Addr2Line(frame->f_code, frame->f_lasti);
#else
    size_t line = 1;
#endif

@hmaarrfk hmaarrfk force-pushed the rebuild-pypy37-0-1_h3a9d69 branch from a1708bc to ac36316 Compare August 20, 2021 14:26
@mattip
Copy link
mattip commented Aug 21, 2021

Thanks, amazingly the build succeeded.
I will try to get these into the next release of PyPy:

@hmaarrfk
Copy link
Contributor

where did the build succeed? i see that i missed a few more const. maybe I'm looking at an old build.

@mattip
Copy link
mattip commented Aug 22, 2021

Ahh, the linking to azure pipelines is broken. Indeed, the pipeline is not passing.

@mattip
Copy link
mattip commented Aug 22, 2021

I am confused why the extra const args are needed for PyPy and not for CPython. It seems to come from pybind11's pytypes.h, but I don't see a const in the sequence_item definition.

@hmaarrfk
Copy link
Contributor

The pytorch source code specifies the inputs as const. So technically, i think the present error is a pytorch bug.

image

@hmaarrfk
Copy link
Contributor

@mattip
Copy link
mattip commented Aug 23, 2021

That test should be skipped for the same reason it is skipped on NumPy due to numpy/numpy#10167. Docstrings are added to the c-extension types by setting tp_doc = ... after PyTypeReady is called. But calling PyTypeReady is the signal in PyPy to create an app-level class object from the PyTypeObject, including rendering the docstring. No further changes to the C structure will be reflected into the python object.

@hmaarrfk
Copy link
Contributor

Well the good thing is that the builds are "green".

The bad thing is that the tests fail.....

@hmaarrfk
Copy link
Contributor

It seems that the tests are skipped on purpose
image

@hmaarrfk hmaarrfk added the bot-rerun Instruct the bot to retry the PR label Jun 8, 2022
@regro-cf-autotick-bot
Copy link
Contributor Author

Due to the bot-rerun label I'm closing this PR. I will make another one as appropriate. This was generated by https://github.com/regro/autotick-bot/actions/runs/2461192229

@regro-cf-autotick-bot regro-cf-autotick-bot deleted the rebuild-pypy37-0-1_h3a9d69 branch June 8, 2022 11:39
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bot-rerun Instruct the bot to retry the PR
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants
0