diff options
| author | Ross Burton <ross.burton@arm.com> | 2025-04-29 16:29:30 +0100 |
|---|---|---|
| committer | Richard Purdie <richard.purdie@linuxfoundation.org> | 2025-05-01 14:22:53 +0100 |
| commit | 83d0d473d6d5a8d59875b0af581d27267e804ba9 (patch) | |
| tree | 52144fc0087e5f8e9836813b2611d3f5d585c76b /meta/recipes-devtools/python/python3/deterministic_imports.patch | |
| parent | cf9c5ac2c152a7b53a1306cbde88bc3c0131028f (diff) | |
| download | poky-83d0d473d6d5a8d59875b0af581d27267e804ba9.tar.gz | |
python3: backport the full fix for importlib scanning invalid distributions
Even with our fixes in deterministic_imports.patch the
importlib.metadata package scan was still returning Distribution objects
for empty directories. This interacts badly with rebuilds when recipes
are changing as when a recipe is removed from the sysroot directories
are not removed[1].
In particular this breaks python3-meson-python-native rebuilds when
Meson upgrades from 1.7 to 1.8: the site-packages directory has an empty
meson-1.7.dist-info/ and populated meson-1.8.dist-info/. Whilst it's
deterministic to return the empty 1.7 first, this breaks pypa/build as
it looks through the distributions in order.
We had discussed this with upstream previously and there's a more
comprehensive fix upstream (actually in importlib_metadata, not cpython)
which ensures that valid distribution objects are listed first. So we
can drop our patch and replace it with a backport to fix these rebuilds.
[1] oe-core 4f94d929639 ("sstate/staging: Handle directory creation race issue")
(From OE-Core rev: 73de8daa6293403f5b92d313af32882c47bce396)
Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Diffstat (limited to 'meta/recipes-devtools/python/python3/deterministic_imports.patch')
| -rw-r--r-- | meta/recipes-devtools/python/python3/deterministic_imports.patch | 39 |
1 files changed, 0 insertions, 39 deletions
diff --git a/meta/recipes-devtools/python/python3/deterministic_imports.patch b/meta/recipes-devtools/python/python3/deterministic_imports.patch deleted file mode 100644 index 61f136ef42..0000000000 --- a/meta/recipes-devtools/python/python3/deterministic_imports.patch +++ /dev/null | |||
| @@ -1,39 +0,0 @@ | |||
| 1 | From 0a02e3b85176a5ce4dd98830bb65dac8596142e9 Mon Sep 17 00:00:00 2001 | ||
| 2 | From: Richard Purdie <richard.purdie@linuxfoundation.org> | ||
| 3 | Date: Fri, 27 May 2022 17:05:44 +0100 | ||
| 4 | Subject: [PATCH] python3: Ensure stale empty python module directories don't | ||
| 5 | |||
| 6 | There are two issues here. Firstly, the modules are accessed in on disk order. This | ||
| 7 | means behaviour seen on one system might not reproduce on another and is a real headache. | ||
| 8 | |||
| 9 | Secondly, empty directories left behind by previous modules might be looked at. This | ||
| 10 | has caused a long string of different issues for us. | ||
| 11 | |||
| 12 | As a result, patch this to a behaviour which works for us. | ||
| 13 | |||
| 14 | Upstream-Status: Submitted [https://github.com/python/cpython/issues/120492; need to first talk to upstream to see if they'll take one or both fixes] | ||
| 15 | Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org> | ||
| 16 | --- | ||
| 17 | Lib/importlib/metadata/__init__.py | 9 ++++++++- | ||
| 18 | 1 file changed, 8 insertions(+), 1 deletion(-) | ||
| 19 | |||
| 20 | diff --git a/Lib/importlib/metadata/__init__.py b/Lib/importlib/metadata/__init__.py | ||
| 21 | index 8ce62dd..a6ea6e9 100644 | ||
| 22 | --- a/Lib/importlib/metadata/__init__.py | ||
| 23 | +++ b/Lib/importlib/metadata/__init__.py | ||
| 24 | @@ -786,7 +786,14 @@ class Lookup: | ||
| 25 | self.infos = FreezableDefaultDict(list) | ||
| 26 | self.eggs = FreezableDefaultDict(list) | ||
| 27 | |||
| 28 | - for child in path.children(): | ||
| 29 | + for child in sorted(path.children()): | ||
| 30 | + childpath = pathlib.Path(path.root, child) | ||
| 31 | + try: | ||
| 32 | + if childpath.is_dir() and not any(childpath.iterdir()): | ||
| 33 | + # Empty directories aren't interesting | ||
| 34 | + continue | ||
| 35 | + except PermissionError: | ||
| 36 | + continue | ||
| 37 | low = child.lower() | ||
| 38 | if low.endswith((".dist-info", ".egg-info")): | ||
| 39 | # rpartition is faster than splitext and suitable for this purpose. | ||
