summaryrefslogtreecommitdiffstats
path: root/meta/recipes-devtools/python/python3/deterministic_imports.patch
diff options
context:
space:
mode:
authorRichard Purdie <richard.purdie@linuxfoundation.org>2022-05-27 17:05:44 +0100
committerRichard Purdie <richard.purdie@linuxfoundation.org>2022-06-24 23:57:42 +0100
commitdaa47a61714a00d7a8e04a521e5185dcde8ca766 (patch)
tree24b59c348dc2bcc08c31071ac62be4e4fd8cd32e /meta/recipes-devtools/python/python3/deterministic_imports.patch
parentd841c3524b2b08ad45f74100059ca912205a2ace (diff)
downloadpoky-daa47a61714a00d7a8e04a521e5185dcde8ca766.tar.gz
python3: Ensure stale empty python module directories don't break the build
There are two issues inside importlib. Firstly, the modules are accessed in on disk order. This means behaviour seen on one system might not reproduce on another and is a real headache. Secondly, empty directories left behind by previous modules might be looked at. This has caused a long string of different issues for us. As a result, patch this to a behaviour which works for us. Upstream discussion can follow later, this is breaking builds for too many people to leave unpatched. [YOCTO #14816] (From OE-Core rev: cdb6879b49d4f6dc2bea8ff064a5b1e62db34781) Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org> (cherry picked from commit e5944a38db513e033c3a3e9313267055f7254be7) Signed-off-by: Steve Sakoman <steve@sakoman.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.patch32
1 files changed, 32 insertions, 0 deletions
diff --git a/meta/recipes-devtools/python/python3/deterministic_imports.patch b/meta/recipes-devtools/python/python3/deterministic_imports.patch
new file mode 100644
index 0000000000..296b413cab
--- /dev/null
+++ b/meta/recipes-devtools/python/python3/deterministic_imports.patch
@@ -0,0 +1,32 @@
1There are two issues here. Firstly, the modules are accessed in on disk order. This
2means behaviour seen on one system might not reproduce on another and is a real headache.
3
4Secondly, empty directories left behind by previous modules might be looked at. This
5has caused a long string of different issues for us.
6
7As a result, patch this to a behaviour which works for us.
8
9Upstream-Status: Pending [need to talk to upstream to see if they'll take one or both fixes]
10Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
11
12
13Index: Python-3.10.4/Lib/importlib/metadata/__init__.py
14===================================================================
15--- Python-3.10.4.orig/Lib/importlib/metadata/__init__.py
16+++ Python-3.10.4/Lib/importlib/metadata/__init__.py
17@@ -819,7 +819,14 @@ class Lookup:
18 self.infos = FreezableDefaultDict(list)
19 self.eggs = FreezableDefaultDict(list)
20
21- for child in path.children():
22+ for child in sorted(path.children()):
23+ childpath = pathlib.Path(path.root, child)
24+ try:
25+ if childpath.is_dir() and not any(childpath.iterdir()):
26+ # Empty directories aren't interesting
27+ continue
28+ except PermissionError:
29+ continue
30 low = child.lower()
31 if low.endswith((".dist-info", ".egg-info")):
32 # rpartition is faster than splitext and suitable for this purpose.