On Tue, Sep 28, 2021 at 9:53 AM Paolo Bonzini <pbonz...@redhat.com> wrote:
> I suspect no one has ever looked at socket_scm_helper.c, because when I
> did my reaction was just "why". The functionality of this 136-line program
> can be reproduced in Python with fewer lines of code than it takes to
> invoke it. Do it, and let it rest in peace.
>
> The only reason not to do that could be because of the upcoming switch
> to aqmp. For that, the new send_fd method has to be wrapped to use
> transport.get_extra_info('socket') and loop.run_in_executor. Let me
> know if you prefer to hold on this until later.
>
> Paolo
>
> Paolo Bonzini (4):
> python: stop using socket_scm_helper
> socket_scm_helper: remove
> python: raise OSError from send_fd_scm
> python: split the two sides of send_fd_scm
>
> python/qemu/machine/machine.py | 58 +++--------
> python/qemu/machine/qtest.py | 2 -
> python/qemu/qmp/__init__.py | 15 +++
> tests/Makefile.include | 5 +-
> tests/meson.build | 4 -
> tests/qemu-iotests/045 | 3 +-
> tests/qemu-iotests/147 | 3 +-
> tests/qemu-iotests/iotests.py | 3 -
> tests/qemu-iotests/meson.build | 5 -
> tests/qemu-iotests/socket_scm_helper.c | 136 -------------------------
> tests/qemu-iotests/testenv.py | 8 +-
> 11 files changed, 35 insertions(+), 207 deletions(-)
> delete mode 100644 tests/qemu-iotests/meson.build
> delete mode 100644 tests/qemu-iotests/socket_scm_helper.c
>
> --
> 2.31.1
>
We discussed this extremely briefly on IRC, but I have some similar
patches already floating around on the list, so I won't be taking these.
Nevertheless, we're in agreement about socket_scm_helper setting sail for
the great beyond!
Any additional cleanups made in these patches I'll just pull into my
ongoing series of Python cleanups and add additional authorship credits.
Thanks!
--js
(For patchew? I don't know if this works. Let's find out:)
NACK
