git: a642f92e86d3 - main - misc/py-pexpect: fix most of the test cases
Po-Chuan Hsieh
sunpoet at freebsd.org
Mon Sep 27 19:48:31 UTC 2021
On Mon, Sep 27, 2021 at 11:50 PM Stefan Eßer <se at freebsd.org> wrote:
> The branch main has been updated by se:
>
> URL:
> https://cgit.FreeBSD.org/ports/commit/?id=a642f92e86d3b022e67d99f1c1dbd9fe6499c634
>
> commit a642f92e86d3b022e67d99f1c1dbd9fe6499c634
> Author: Stefan Eßer <se at FreeBSD.org>
> AuthorDate: 2021-09-27 13:49:11 +0000
> Commit: Stefan Eßer <se at FreeBSD.org>
> CommitDate: 2021-09-27 15:49:50 +0000
>
> misc/py-pexpect: fix most of the test cases
>
> A number of test cases included in the pexpect distribution failed on
> FreeBSD. The patches included in the commit fix most of these tests.
>
> One of the tests used /bin/bash and long options for the nl command.
> Other tests assumed pipe buffers larger than provided by FreeBSD.
>
> 3 tests are still failing:
>
> tests/test_replwrap.py::REPLWrapTestCase::test_existing_spawn FAILED
>
> The text received contains bracketed paste escape characters when
> a bash version 5.1 or newer is installed. There may be another problem
> with this test, though, since it seems to depend on some bash rc file
> of the account executing this test.
>
> tests/test_socket.py::ExpectTestCase::test_interrupt FAILED
> tests/test_socket.py::ExpectTestCase::test_multiple_interrupts FAILED
>
> These two test seem to fail due to an early exit of the sending
> process. It appears to be in Zombie state and cannot be signaled.
> The "non-blocking read" seems to be blocking in this situation.
>
> The real cause is unclear, it seems to be a mismatch between process
> lifetime semantics in Linux and FreeBSD. I have removed an infinite
> wait loop with these patches, which causes a test failure to be
> detected (instead of preventing the execution of any further test
> cases).
>
> I have created a pull request on Github to get these patches into the
> upstream, but it seems that the owner of the pexpect repository is not
> very active.
>
> Approved by: sunpoet (maintainer)
>
Stefan, thanks for fixing it!
Regards,
sunpoet
More information about the dev-commits-ports-all
mailing list