| Message ID | 20240813164101.3425564-1-tgamblin@baylibre.com |
|---|---|
| State | Accepted, archived |
| Commit | de569ddffd5ea36b70c56df21dec9c892e5dee7d |
| Headers | show |
| Series | [v2] python3: skip readline limited history tests | expand |
The github issue you referenced here is not an issue that tracks this
test failure. It's an issue that introduced the original commit, and has
been closed. If a problem is 100% reproducible, at least some open issue
should be there for tracking to validate an explicit skip. As far as I
can see, tests that were skipped are rarely revisited.
Also, looking at the issue you referenced, it seems that the problem is
that we're using editline instead of readline as the default
PACKAGECONFIG. It looks like that this new test case correctly reveals
our specific configuration problem.
PACKAGECONFIG ??= "editline gdbm ${@bb.utils.filter('DISTRO_FEATURES',
'lto', d)}"
PACKAGECONFIG[readline] = "--with-readline=readline,,readline,,,editline"
PACKAGECONFIG[editline] = "--with-readline=editline,,libedit,,,readline"
Can you try using readline as the default PACKAGECONFIG and see if
things work?
Regards,
Qi
On 8/14/24 00:41, Trevor Gamblin wrote:
> Python 3.12.5 is failing a newer ptest for reading/writing limited
> history. Skip it for now until a proper fix (if any) is determined.
>
> The new test was added in
> https://github.com/python/cpython/commit/263c7e611bb24715e513d457a3477a61fff15162
>
> The GitHub issue is here: https://github.com/python/cpython/issues/121160
>
> The aforementioned bug report suggests that OSErrors like that show up
> when using the GNU readline-based readline module if a workaround isn't
> also incorporated. It also mentions a new readline backend will be
> available in 3.13 that might help, but that won't be available for a year.
>
> Signed-off-by: Trevor Gamblin <tgamblin@baylibre.com>
> ---
> v2 adds extra info in the commit message and carried patch about the
> source of the test failure.
>
> ...t_readline-skip-limited-history-test.patch | 40 +++++++++++++++++++
> .../recipes-devtools/python/python3_3.12.5.bb | 1 +
> 2 files changed, 41 insertions(+)
> create mode 100644 meta/recipes-devtools/python/python3/0001-test_readline-skip-limited-history-test.patch
>
> diff --git a/meta/recipes-devtools/python/python3/0001-test_readline-skip-limited-history-test.patch b/meta/recipes-devtools/python/python3/0001-test_readline-skip-limited-history-test.patch
> new file mode 100644
> index 0000000000..851ddd571a
> --- /dev/null
> +++ b/meta/recipes-devtools/python/python3/0001-test_readline-skip-limited-history-test.patch
> @@ -0,0 +1,40 @@
> +From d9d916d5ea946c945323679d1709de1b87029b96 Mon Sep 17 00:00:00 2001
> +From: Trevor Gamblin <tgamblin@baylibre.com>
> +Date: Tue, 13 Aug 2024 11:07:05 -0400
> +Subject: [PATCH] test_readline: skip limited history test
> +
> +This test was added recently and is failing on the ptest image. Disable
> +it until the proper fix is determined.
> +
> +See also: https://github.com/python/cpython/issues/121160
> +
> +Upstream-Status: Inappropriate [OE-specific]
> +
> +Signed-off-by: Trevor Gamblin <tgamblin@baylibre.com>
> +---
> + Lib/test/test_readline.py | 2 ++
> + 1 file changed, 2 insertions(+)
> +
> +diff --git a/Lib/test/test_readline.py b/Lib/test/test_readline.py
> +index 91fd7dd13f9..d81f9bf8eed 100644
> +--- a/Lib/test/test_readline.py
> ++++ b/Lib/test/test_readline.py
> +@@ -132,6 +132,7 @@ def test_nonascii_history(self):
> + self.assertEqual(readline.get_history_item(1), "entrée 1")
> + self.assertEqual(readline.get_history_item(2), "entrée 22")
> +
> ++ @unittest.skip("Skipping problematic test")
> + def test_write_read_limited_history(self):
> + previous_length = readline.get_history_length()
> + self.addCleanup(readline.set_history_length, previous_length)
> +@@ -349,6 +350,7 @@ def test_history_size(self):
> + self.assertEqual(len(lines), history_size)
> + self.assertEqual(lines[-1].strip(), b"last input")
> +
> ++ @unittest.skip("Skipping problematic test")
> + def test_write_read_limited_history(self):
> + previous_length = readline.get_history_length()
> + self.addCleanup(readline.set_history_length, previous_length)
> +--
> +2.39.2
> +
> diff --git a/meta/recipes-devtools/python/python3_3.12.5.bb b/meta/recipes-devtools/python/python3_3.12.5.bb
> index 5665c90fa9..cc2e14baf0 100644
> --- a/meta/recipes-devtools/python/python3_3.12.5.bb
> +++ b/meta/recipes-devtools/python/python3_3.12.5.bb
> @@ -33,6 +33,7 @@ SRC_URI = "http://www.python.org/ftp/python/${PV}/Python-${PV}.tar.xz \
> file://0001-gh-107811-tarfile-treat-overflow-in-UID-GID-as-failu.patch \
> file://0001-test_deadlock-skip-problematic-test.patch \
> file://0001-test_active_children-skip-problematic-test.patch \
> + file://0001-test_readline-skip-limited-history-test.patch \
> "
>
> SRC_URI:append:class-native = " \
>
> -=-=-=-=-=-=-=-=-=-=-=-
> Links: You receive all messages sent to this group.
> View/Reply Online (#203289): https://lists.openembedded.org/g/openembedded-core/message/203289
> Mute This Topic: https://lists.openembedded.org/mt/107879487/7304865
> Group Owner: openembedded-core+owner@lists.openembedded.org
> Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [Qi.Chen@eng.windriver.com]
> -=-=-=-=-=-=-=-=-=-=-=-
>
On 2024-08-13 10:56 p.m., ChenQi wrote: > The github issue you referenced here is not an issue that tracks this > test failure. It's an issue that introduced the original commit, and > has been closed. If a problem is 100% reproducible, at least some open > issue should be there for tracking to validate an explicit skip. As > far as I can see, tests that were skipped are rarely revisited. > > Also, looking at the issue you referenced, it seems that the problem > is that we're using editline instead of readline as the default > PACKAGECONFIG. It looks like that this new test case correctly reveals > our specific configuration problem. > PACKAGECONFIG ??= "editline gdbm ${@bb.utils.filter('DISTRO_FEATURES', > 'lto', d)}" > PACKAGECONFIG[readline] = "--with-readline=readline,,readline,,,editline" > PACKAGECONFIG[editline] = "--with-readline=editline,,libedit,,,readline" > > Can you try using readline as the default PACKAGECONFIG and see if > things work? Yes, the test passes with readline. I didn't make this change for the test because I had assumed we want to use editline in the PACKAGECONFIG instead of readline, based on commit ce6add6f8aca0. Is this not the case? Do we want to switch back to using readline by default? > > Regards, > Qi > > On 8/14/24 00:41, Trevor Gamblin wrote: >> Python 3.12.5 is failing a newer ptest for reading/writing limited >> history. Skip it for now until a proper fix (if any) is determined. >> >> The new test was added in >> https://github.com/python/cpython/commit/263c7e611bb24715e513d457a3477a61fff15162 >> >> >> The GitHub issue is here: >> https://github.com/python/cpython/issues/121160 >> >> The aforementioned bug report suggests that OSErrors like that show up >> when using the GNU readline-based readline module if a workaround isn't >> also incorporated. It also mentions a new readline backend will be >> available in 3.13 that might help, but that won't be available for a >> year. >> >> Signed-off-by: Trevor Gamblin <tgamblin@baylibre.com> >> --- >> v2 adds extra info in the commit message and carried patch about the >> source of the test failure. >> >> ...t_readline-skip-limited-history-test.patch | 40 +++++++++++++++++++ >> .../recipes-devtools/python/python3_3.12.5.bb | 1 + >> 2 files changed, 41 insertions(+) >> create mode 100644 >> meta/recipes-devtools/python/python3/0001-test_readline-skip-limited-history-test.patch >> >> diff --git >> a/meta/recipes-devtools/python/python3/0001-test_readline-skip-limited-history-test.patch >> b/meta/recipes-devtools/python/python3/0001-test_readline-skip-limited-history-test.patch >> >> new file mode 100644 >> index 0000000000..851ddd571a >> --- /dev/null >> +++ >> b/meta/recipes-devtools/python/python3/0001-test_readline-skip-limited-history-test.patch >> @@ -0,0 +1,40 @@ >> +From d9d916d5ea946c945323679d1709de1b87029b96 Mon Sep 17 00:00:00 2001 >> +From: Trevor Gamblin <tgamblin@baylibre.com> >> +Date: Tue, 13 Aug 2024 11:07:05 -0400 >> +Subject: [PATCH] test_readline: skip limited history test >> + >> +This test was added recently and is failing on the ptest image. Disable >> +it until the proper fix is determined. >> + >> +See also: https://github.com/python/cpython/issues/121160 >> + >> +Upstream-Status: Inappropriate [OE-specific] >> + >> +Signed-off-by: Trevor Gamblin <tgamblin@baylibre.com> >> +--- >> + Lib/test/test_readline.py | 2 ++ >> + 1 file changed, 2 insertions(+) >> + >> +diff --git a/Lib/test/test_readline.py b/Lib/test/test_readline.py >> +index 91fd7dd13f9..d81f9bf8eed 100644 >> +--- a/Lib/test/test_readline.py >> ++++ b/Lib/test/test_readline.py >> +@@ -132,6 +132,7 @@ def test_nonascii_history(self): >> + self.assertEqual(readline.get_history_item(1), "entrée 1") >> + self.assertEqual(readline.get_history_item(2), "entrée 22") >> + >> ++ @unittest.skip("Skipping problematic test") >> + def test_write_read_limited_history(self): >> + previous_length = readline.get_history_length() >> + self.addCleanup(readline.set_history_length, previous_length) >> +@@ -349,6 +350,7 @@ def test_history_size(self): >> + self.assertEqual(len(lines), history_size) >> + self.assertEqual(lines[-1].strip(), b"last input") >> + >> ++ @unittest.skip("Skipping problematic test") >> + def test_write_read_limited_history(self): >> + previous_length = readline.get_history_length() >> + self.addCleanup(readline.set_history_length, previous_length) >> +-- >> +2.39.2 >> + >> diff --git a/meta/recipes-devtools/python/python3_3.12.5.bb >> b/meta/recipes-devtools/python/python3_3.12.5.bb >> index 5665c90fa9..cc2e14baf0 100644 >> --- a/meta/recipes-devtools/python/python3_3.12.5.bb >> +++ b/meta/recipes-devtools/python/python3_3.12.5.bb >> @@ -33,6 +33,7 @@ SRC_URI = >> "http://www.python.org/ftp/python/${PV}/Python-${PV}.tar.xz \ >> file://0001-gh-107811-tarfile-treat-overflow-in-UID-GID-as-failu.patch \ >> file://0001-test_deadlock-skip-problematic-test.patch \ >> file://0001-test_active_children-skip-problematic-test.patch \ >> + file://0001-test_readline-skip-limited-history-test.patch \ >> " >> SRC_URI:append:class-native = " \ >> >> -=-=-=-=-=-=-=-=-=-=-=- >> Links: You receive all messages sent to this group. >> View/Reply Online (#203289): >> https://lists.openembedded.org/g/openembedded-core/message/203289 >> Mute This Topic: https://lists.openembedded.org/mt/107879487/7304865 >> Group Owner: openembedded-core+owner@lists.openembedded.org >> Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub >> [Qi.Chen@eng.windriver.com] >> -=-=-=-=-=-=-=-=-=-=-=- >> >
How about using INCOMPITABLE_LICENSE to determine the default PACAKGECONFIG, so that the default is readline? Maybe this would be part of the new meta-gplv2 replacement work (@Khem Raj). I'm not very sure about it, though. As far as I can see, very few people care about this GPLv3 problem. The evidence is that meta-gplv2 is NOT supported for a long time and nobody ever devotes any real effort to develop an alternative. I guess when the commit was made, the underlying API incompatibility was NOT made clear. I'm OK with any choice, as long as clear statements are there to tell people why, though I prefer to use readline as the default, as it would provide better user experience. Regards, Qi -----Original Message----- From: Trevor Gamblin <tgamblin@baylibre.com> Sent: Wednesday, August 14, 2024 9:04 PM To: Chen, Qi <Qi.Chen@windriver.com>; openembedded-core@lists.openembedded.org Cc: alex.kanavin@gmail.com; alexandre.belloni@bootlin.com Subject: Re: [OE-core][PATCH v2] python3: skip readline limited history tests On 2024-08-13 10:56 p.m., ChenQi wrote: > The github issue you referenced here is not an issue that tracks this > test failure. It's an issue that introduced the original commit, and > has been closed. If a problem is 100% reproducible, at least some open > issue should be there for tracking to validate an explicit skip. As > far as I can see, tests that were skipped are rarely revisited. > > Also, looking at the issue you referenced, it seems that the problem > is that we're using editline instead of readline as the default > PACKAGECONFIG. It looks like that this new test case correctly reveals > our specific configuration problem. > PACKAGECONFIG ??= "editline gdbm ${@bb.utils.filter('DISTRO_FEATURES', > 'lto', d)}" > PACKAGECONFIG[readline] = "--with-readline=readline,,readline,,,editline" > PACKAGECONFIG[editline] = "--with-readline=editline,,libedit,,,readline" > > Can you try using readline as the default PACKAGECONFIG and see if > things work? Yes, the test passes with readline. I didn't make this change for the test because I had assumed we want to use editline in the PACKAGECONFIG instead of readline, based on commit ce6add6f8aca0. Is this not the case? Do we want to switch back to using readline by default? > > Regards, > Qi > > On 8/14/24 00:41, Trevor Gamblin wrote: >> Python 3.12.5 is failing a newer ptest for reading/writing limited >> history. Skip it for now until a proper fix (if any) is determined. >> >> The new test was added in >> https://github.com/python/cpython/commit/263c7e611bb24715e513d457a347 >> 7a61fff15162 >> >> >> The GitHub issue is here: >> https://github.com/python/cpython/issues/121160 >> >> The aforementioned bug report suggests that OSErrors like that show >> up when using the GNU readline-based readline module if a workaround >> isn't also incorporated. It also mentions a new readline backend will >> be available in 3.13 that might help, but that won't be available for >> a year. >> >> Signed-off-by: Trevor Gamblin <tgamblin@baylibre.com> >> --- >> v2 adds extra info in the commit message and carried patch about the >> source of the test failure. >> >> ...t_readline-skip-limited-history-test.patch | 40 >> +++++++++++++++++++ >> .../recipes-devtools/python/python3_3.12.5.bb | 1 + >> 2 files changed, 41 insertions(+) >> create mode 100644 >> meta/recipes-devtools/python/python3/0001-test_readline-skip-limited- >> history-test.patch >> >> diff --git >> a/meta/recipes-devtools/python/python3/0001-test_readline-skip-limite >> d-history-test.patch >> b/meta/recipes-devtools/python/python3/0001-test_readline-skip-limite >> d-history-test.patch >> >> new file mode 100644 >> index 0000000000..851ddd571a >> --- /dev/null >> +++ >> b/meta/recipes-devtools/python/python3/0001-test_readline-skip-limite >> d-history-test.patch >> @@ -0,0 +1,40 @@ >> +From d9d916d5ea946c945323679d1709de1b87029b96 Mon Sep 17 00:00:00 >> +2001 >> +From: Trevor Gamblin <tgamblin@baylibre.com> >> +Date: Tue, 13 Aug 2024 11:07:05 -0400 >> +Subject: [PATCH] test_readline: skip limited history test >> + >> +This test was added recently and is failing on the ptest image. >> +Disable it until the proper fix is determined. >> + >> +See also: https://github.com/python/cpython/issues/121160 >> + >> +Upstream-Status: Inappropriate [OE-specific] >> + >> +Signed-off-by: Trevor Gamblin <tgamblin@baylibre.com> >> +--- >> + Lib/test/test_readline.py | 2 ++ >> + 1 file changed, 2 insertions(+) >> + >> +diff --git a/Lib/test/test_readline.py b/Lib/test/test_readline.py >> +index 91fd7dd13f9..d81f9bf8eed 100644 >> +--- a/Lib/test/test_readline.py >> ++++ b/Lib/test/test_readline.py >> +@@ -132,6 +132,7 @@ def test_nonascii_history(self): >> + self.assertEqual(readline.get_history_item(1), "entrée 1") >> + self.assertEqual(readline.get_history_item(2), "entrée 22") >> + >> ++ @unittest.skip("Skipping problematic test") >> + def test_write_read_limited_history(self): >> + previous_length = readline.get_history_length() >> + self.addCleanup(readline.set_history_length, >> +previous_length) @@ -349,6 +350,7 @@ def test_history_size(self): >> + self.assertEqual(len(lines), history_size) >> + self.assertEqual(lines[-1].strip(), b"last input") >> + >> ++ @unittest.skip("Skipping problematic test") >> + def test_write_read_limited_history(self): >> + previous_length = readline.get_history_length() >> + self.addCleanup(readline.set_history_length, >> +previous_length) >> +-- >> +2.39.2 >> + >> diff --git a/meta/recipes-devtools/python/python3_3.12.5.bb >> b/meta/recipes-devtools/python/python3_3.12.5.bb >> index 5665c90fa9..cc2e14baf0 100644 >> --- a/meta/recipes-devtools/python/python3_3.12.5.bb >> +++ b/meta/recipes-devtools/python/python3_3.12.5.bb >> @@ -33,6 +33,7 @@ SRC_URI = >> "http://www.python.org/ftp/python/${PV}/Python-${PV}.tar.xz \ >> file://0001-gh-107811-tarfile-treat-overflow-in-UID-GID-as-failu.patc >> h \ >> file://0001-test_deadlock-skip-problematic-test.patch \ >> file://0001-test_active_children-skip-problematic-test.patch \ >> + file://0001-test_readline-skip-limited-history-test.patch \ >> " >> SRC_URI:append:class-native = " \ >> >> -=-=-=-=-=-=-=-=-=-=-=- >> Links: You receive all messages sent to this group. >> View/Reply Online (#203289): >> https://lists.openembedded.org/g/openembedded-core/message/203289 >> Mute This Topic: https://lists.openembedded.org/mt/107879487/7304865 >> Group Owner: openembedded-core+owner@lists.openembedded.org >> Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub >> [Qi.Chen@eng.windriver.com] >> -=-=-=-=-=-=-=-=-=-=-=- >> >
> -----Original Message----- > From: openembedded-core@lists.openembedded.org <openembedded- > core@lists.openembedded.org> On Behalf Of Chen Qi via > lists.openembedded.org > Sent: den 14 augusti 2024 17:02 > To: Trevor Gamblin <tgamblin@baylibre.com>; openembedded- > core@lists.openembedded.org; Khem Raj <raj.khem@gmail.com> > Cc: alex.kanavin@gmail.com; alexandre.belloni@bootlin.com; Khem Raj > <raj.khem@gmail.com> > Subject: Re: [OE-core][PATCH v2] python3: skip readline limited history > tests > > How about using INCOMPITABLE_LICENSE to determine the default > PACAKGECONFIG, so that the default is readline? Maybe this would be part No. In modern configurations, INCOMPITABLE_LICENSE should only be set for image recipes to allow, e.g., GPL-3.0 code to be used in one image, but not another. > of the new meta-gplv2 replacement work (@Khem Raj). I'm not very sure > about it, though. As far as I can see, very few people care about this > GPLv3 problem. The evidence is that meta-gplv2 is NOT supported for a long Note that readline is GPL-3.0, not LGPL-3.0, which means linking with it directly impacts the licensing of the entire application. And while I cannot speak for others, (L)GPL-3.0 are problematic and should be avoided if possible, and be optional otherwise. Which is exactly what Ross did when he changed the default from readline to editline in the python3 recipe. Going back on that is definitely the wrong thing to do. > time and nobody ever devotes any real effort to develop an alternative. > > I guess when the commit was made, the underlying API incompatibility was > NOT made clear. If the test fails when python3 is configured to use editline, then I would say that it is the test that is wrong and should be corrected. > > I'm OK with any choice, as long as clear statements are there to tell > people why, though I prefer to use readline as the default, as it would > provide better user experience. > > Regards, > Qi //Peter
On Wed, 14 Aug 2024 at 17:32, Peter Kjellerstedt <peter.kjellerstedt@axis.com> wrote: > Note that readline is GPL-3.0, not LGPL-3.0, which means linking with it > directly impacts the licensing of the entire application. > > And while I cannot speak for others, (L)GPL-3.0 are problematic and should be > avoided if possible, and be optional otherwise. Which is exactly what Ross > did when he changed the default from readline to editline in the python3 > recipe. Going back on that is definitely the wrong thing to do. As much as I think that what GPLv3 aims to do is just, and its terms serve the interest of the end users and general public (protecting their essential freedoms from tyranny of the software vendors), I also have to agree with the above. When there is a choice given to us by upstream, we should go with non-gpl3 options if they don't result in big loss of functionality. > > time and nobody ever devotes any real effort to develop an alternative. > > > > I guess when the commit was made, the underlying API incompatibility was > > NOT made clear. > > If the test fails when python3 is configured to use editline, then I would > say that it is the test that is wrong and should be corrected. Also this is correct. Python's test suite should pass regardless of how it was configured. I'd suggest doing these things: - clearly marking the reason for skipping the test as being readline-specific and failing with editline - filing an upstream bug specifically about this, and including the link to it into the patch. It's pretty much expected nowadays that any new Inappropriate patch should come with such a bug link. Alex
diff --git a/meta/recipes-devtools/python/python3/0001-test_readline-skip-limited-history-test.patch b/meta/recipes-devtools/python/python3/0001-test_readline-skip-limited-history-test.patch new file mode 100644 index 0000000000..851ddd571a --- /dev/null +++ b/meta/recipes-devtools/python/python3/0001-test_readline-skip-limited-history-test.patch @@ -0,0 +1,40 @@ +From d9d916d5ea946c945323679d1709de1b87029b96 Mon Sep 17 00:00:00 2001 +From: Trevor Gamblin <tgamblin@baylibre.com> +Date: Tue, 13 Aug 2024 11:07:05 -0400 +Subject: [PATCH] test_readline: skip limited history test + +This test was added recently and is failing on the ptest image. Disable +it until the proper fix is determined. + +See also: https://github.com/python/cpython/issues/121160 + +Upstream-Status: Inappropriate [OE-specific] + +Signed-off-by: Trevor Gamblin <tgamblin@baylibre.com> +--- + Lib/test/test_readline.py | 2 ++ + 1 file changed, 2 insertions(+) + +diff --git a/Lib/test/test_readline.py b/Lib/test/test_readline.py +index 91fd7dd13f9..d81f9bf8eed 100644 +--- a/Lib/test/test_readline.py ++++ b/Lib/test/test_readline.py +@@ -132,6 +132,7 @@ def test_nonascii_history(self): + self.assertEqual(readline.get_history_item(1), "entrée 1") + self.assertEqual(readline.get_history_item(2), "entrée 22") + ++ @unittest.skip("Skipping problematic test") + def test_write_read_limited_history(self): + previous_length = readline.get_history_length() + self.addCleanup(readline.set_history_length, previous_length) +@@ -349,6 +350,7 @@ def test_history_size(self): + self.assertEqual(len(lines), history_size) + self.assertEqual(lines[-1].strip(), b"last input") + ++ @unittest.skip("Skipping problematic test") + def test_write_read_limited_history(self): + previous_length = readline.get_history_length() + self.addCleanup(readline.set_history_length, previous_length) +-- +2.39.2 + diff --git a/meta/recipes-devtools/python/python3_3.12.5.bb b/meta/recipes-devtools/python/python3_3.12.5.bb index 5665c90fa9..cc2e14baf0 100644 --- a/meta/recipes-devtools/python/python3_3.12.5.bb +++ b/meta/recipes-devtools/python/python3_3.12.5.bb @@ -33,6 +33,7 @@ SRC_URI = "http://www.python.org/ftp/python/${PV}/Python-${PV}.tar.xz \ file://0001-gh-107811-tarfile-treat-overflow-in-UID-GID-as-failu.patch \ file://0001-test_deadlock-skip-problematic-test.patch \ file://0001-test_active_children-skip-problematic-test.patch \ + file://0001-test_readline-skip-limited-history-test.patch \ " SRC_URI:append:class-native = " \
Python 3.12.5 is failing a newer ptest for reading/writing limited history. Skip it for now until a proper fix (if any) is determined. The new test was added in https://github.com/python/cpython/commit/263c7e611bb24715e513d457a3477a61fff15162 The GitHub issue is here: https://github.com/python/cpython/issues/121160 The aforementioned bug report suggests that OSErrors like that show up when using the GNU readline-based readline module if a workaround isn't also incorporated. It also mentions a new readline backend will be available in 3.13 that might help, but that won't be available for a year. Signed-off-by: Trevor Gamblin <tgamblin@baylibre.com> --- v2 adds extra info in the commit message and carried patch about the source of the test failure. ...t_readline-skip-limited-history-test.patch | 40 +++++++++++++++++++ .../recipes-devtools/python/python3_3.12.5.bb | 1 + 2 files changed, 41 insertions(+) create mode 100644 meta/recipes-devtools/python/python3/0001-test_readline-skip-limited-history-test.patch