Currently, test_gdb fails if gdb logs messages on stderr which are "unexpected". I don't understand the rationale for that: Python is not supposed to test gdb. It's only supposed to check that python-gdb.py commands work as expected: stderr should be ignored.

In the past, I was lazy and just added more and more patterns to ignore on stderr, but this approach doesn't work in the long term: gdb evolves frequently, and there are always new messages.

Attached PR modify test_gdb to ignore stderr, except of "PC not saved" pattern used to skip test_gdb on a special case: bpo-34007.

        # bpo34007: Sometimes some versions of the shared libraries that
        # are part of the traceback are compiled in optimised mode and the
        # Program Counter (PC) is not present, not allowing gdb to walk the
        # frames back. When this happens, the Python bindings of gdb raise
        # an exception, making the test impossible to succeed.
        if "PC not saved" in err:
            raise unittest.SkipTest("gdb cannot walk the frame object"
                                    " because the Program Counter is"
                                    " not present")

In the past, I was lazy and just added more and more patterns to ignore on stderr, but this approach doesn't work in the long term: gdb evolves frequently, and there are always new messages.

A recent example: test_gdb fails on Fedora because a warning which should not prevent to test python-gdb.py commands:

"Missing separate debuginfo for /lib/ld-linux-aarch64.so.1"

Moreover, Fedora also suggests a command to install missing package in this case:

"Try: dnf --enablerepo='*debug*' install ..."

https://bugzilla.redhat.com/show_bug.cgi?id=1721483

I think when I wrote this I was over-optimistically thinking that we could just add more patterns, but if it's becoming a pain, then your approach looks good to me.

New changeset e56a123 by Victor Stinner in branch 'master':
bpo-37362: test_gdb now ignores stderr (GH-14287)
e56a123

I think when I wrote this I was over-optimistically thinking that we could just add more patterns, but if it's becoming a pain, then your approach looks good to me.

Well, I tried hard to fit into this approach: over the years, I added more and more patterns... but it's a painful work: first a CI break, I add more patterns and then I have to backport the change to all branches. As I wrote, I'm not convinced of the purpose of getting a failure in this case. Python doesn't get any benefit from this.

Anyway, I merged my change. Thanks for your approval ;-)

New changeset 16ec95b by Miss Islington (bot) in branch '3.7':
bpo-37362: test_gdb now ignores stderr (GH-14287)
16ec95b

New changeset 3523e0c by Miss Islington (bot) in branch '3.8':
bpo-37362: test_gdb now ignores stderr (GH-14287)
3523e0c

New changeset adcdb1e by Victor Stinner in branch '2.7':
bpo-37362: test_gdb now ignores stderr (GH-14287) (GH-14297)
adcdb1e

For the record, examples of ignored patterns:

        ignore_patterns = (
            'Function "%s" not defined.' % breakpoint,
            'Do you need "set solib-search-path" or '
            '"set sysroot"?',
            # BFD: /usr/lib/debug/(...): unable to initialize decompress
            # status for section .debug_aranges
            'BFD: ',
            # ignore all warnings
            'warning: ',
            )

Enjoy the older list before I chose to ignore "warning: " :-)

        # Ignore some benign messages on stderr.
        ignore_patterns = (
            'Function "%s" not defined.' % breakpoint,
            "warning: no loadable sections found in added symbol-file"
            " system-supplied DSO",
            "warning: Unable to find libthread_db matching"
            " inferior's thread library, thread debugging will"
            " not be available.",
            "warning: Cannot initialize thread debugging"
            " library: Debugger service failed",
            'warning: Could not load shared library symbols for '
            'linux-vdso.so',
            'warning: Could not load shared library symbols for '
            'linux-gate.so',
            'warning: Could not load shared library symbols for '
            'linux-vdso64.so',
            'Do you need "set solib-search-path" or '
            '"set sysroot"?',
            'warning: Source file is more recent than executable.',
            # Issue python/cpython#63952: missing symbols on System Z
            'Missing separate debuginfo for ',
            'Try: zypper install -C ',
            )

Oh strange, "Missing separate debuginfo for " message was ignored!

commit f4a4898
Author: Victor Stinner <victor.stinner@gmail.com>
Date: Sun Nov 24 18:55:25 2013 +0100

Issue bpo-19753: Try to fix test_gdb on SystemZ buildbot

But I removed it in:

commit 904f5de
Author: Victor Stinner <victor.stinner@gmail.com>
Date: Wed Mar 23 18:32:54 2016 +0100

Try to fix test_gdb on s390x buildbots

I closed the issue. test_gdb now ignores stderr in all branches.