| Patch | Description | Author | Forwarded | Bugs | Origin | Last update |
|---|---|---|---|---|---|---|
| dirtyhack.patch | The elegant patching work is based on the master branch https://github.com/pytorch/pytorch/issues/14699 And we will be able to use that solution in the next upstream release. I don't want to rebase my patches back to this version, so let's go with a fast, yet dirty hack. =================================================================== |
Mo Zhou | no | |||
| mkldnn.patch | =================================================================== | no | ||||
| zstd.patch | =================================================================== | no | ||||
| cmake-strip-3rdparty.patch | =================================================================== | no | ||||
| shebang.patch | change shebang find . -type f -name '*.py' -exec sed -i -e 's@#!/usr/bin/env @#!/usr/bin/@g' '{}' \; find . -type f -name '*.sh' -exec sed -i -e 's@#!/usr/bin/env @#!/usr/bin/@g' '{}' \; git diff > x.shebang cat x.shebang >> debian/patches/shebang.patch =================================================================== |
Mo Zhou | no | |||
| pytorch_glog_update.patch | move IsGoogleLoggingInitialized() to public API It was an internal function and project used hacks to reach it. Now it's part of the public API. |
Laszlo Boszormenyi (GCS) <gcs@debian.org> | no | 2022-03-08 | ||
| flatbuffers-v2.0.8.patch | flatbuffers API breakage fix in debian/rules, you can find a line invoking `flatc` to generate some cpp code from flatbuffers protocol file. The pytorch upstream code is based on some early version of flatbuffers, while our flatbuffers version (v2.0.8) is much newer than that supported by pytorch upstream. As a result, there are some API mismatches. This patch is to fix these API mismatch and hence fix FTBFS. But the upstream (seemingly) does not intent to bump their dependency lib. =================================================================== |
Mo Zhou | yes | |||
| fix-wrong-shebang.patch | fix wrong shebang=================================================================== | no | ||||
| fmtlib-revert.patch | revert the string formatting overhead. Basically a partial revert of this: https://github.com/pytorch/pytorch/pull/76977 we encountered a strange FTBFS issue when compiling against libfmt. reverting back to c++ std and try again. This patch is not yet verified. diff --git a/torch/csrc/Exceptions.cpp b/torch/csrc/Exceptions.cpp index 5210d6f7..d0dfbc89 100644 |
no | ||||
| 0009-Fix-corner-cases-with-permute-88226.patch | Fix corner cases with permute (#88226) Previously the permute function was extended to behave like the `order` function for first-class dimensions. However, unlike `permute`, `order` doesn't have a keyword argment `dims`, and there is no way to add it in a way that makes both permute an order to continue to have the same behavior. So this change just removes the extra functionality of permute, which wasn't documented anyway. Fixes #88187 Pull Request resolved: https://github.com/pytorch/pytorch/pull/88226 Approved by: https://github.com/zou3519 |
Zachary DeVito <zdevito@meta.com> | no | 2022-11-01 | ||
| 0010-fix-possible-overflow-83389.patch | fix possible overflow (#83389) Fix some errors detected by static analysis. Pull Request resolved: https://github.com/pytorch/pytorch/pull/83389 Approved by: https://github.com/zou3519 |
cyy <cyyever@outlook.com> | no | 2022-11-29 | ||
| 0011-Use-python-compat-from-python-pythoncapi_compat-9116.patch | Use python compat from python/pythoncapi_compat (#91163) Pull Request resolved: https://github.com/pytorch/pytorch/pull/91163 Approved by: https://github.com/ezyang |
albanD <desmaison.alban@gmail.com> | no | 2022-12-21 | ||
| 0013-Fix-test_math_ops-for-python-3.11-91774.patch | Fix `test_math_ops` for python-3.11 (#91774) From [math.pow](https://docs.python.org/3/library/math.html#math.pow) documentation: > Changed in version 3.11: The special cases `pow(0.0, -inf)` and `pow(-0.0, -inf)` were changed to return `inf` instead of raising [`ValueError`](https://docs.python.org/3/library/exceptions.html#ValueError), for consistency with IEEE 754. Pull Request resolved: https://github.com/pytorch/pytorch/pull/91774 Approved by: https://github.com/ngimel |
Nikita Shulga <nshulga@meta.com> | no | 2023-01-06 | ||
| 0015-Add-missing-gc-untrack-for-cpp-autograd-Nodes-92351.patch | Add missing gc untrack for cpp autograd Nodes (#92351) Fixes https://github.com/pytorch/pytorch/issues/91161 the assertion after the warning seems to be linked to the fact that we didn't untrack this properly. In 3.11 they added a warning when this is not called properly before tp_free Pull Request resolved: https://github.com/pytorch/pytorch/pull/92351 Approved by: https://github.com/ezyang |
albanD <desmaison.alban@gmail.com> | no | 2023-01-18 | ||
| 0014-Skip-builtins-while-enumerating-class-methods-91805.patch | Skip builtins while enumerating class methods (#91805) This is needed to support `enum.Enum` derived classes in Python-3.11 that adds `_new_member_` to classdict, see: https://github.com/python/cpython/blob/15c44789bb125b93e96815a336ec73423c47508e/Lib/enum.py#L529 Following snippet illustrates the problem with the previous iteration of the code on 3.11: ```python from enum import Enum import inspect class Color(Enum): RED = 1 GREEN = 2 def print_routines(cls): print(cls.__name__) for name in cls.__dict__: fn = getattr(cls, name) if inspect.isroutine(fn): print(name, fn, f"has_globals: {hasattr(fn, '__globals__')}") print_routines(Color) ``` Pull Request resolved: https://github.com/pytorch/pytorch/pull/91805 Approved by: https://github.com/albanD, https://github.com/suo |
Nikita Shulga <nshulga@meta.com> | no | 2023-01-06 |