Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 04 Oct 2023 12:16:36 -0300
From:      Joseph Mingrone <jrm@FreeBSD.org>
To:        freebsd-python@FreeBSD.org
Subject:   [Victor Stinner] Re: Python buildbots for FreeBSD
Message-ID:  <868r8ia04r.fsf@phe.ftfl.ca>

next in thread | raw e-mail | index | archive | help
--=-=-=
Content-Type: text/plain

Hello,

Some people recently added new Python buildbots for FreeBSD.  Victor
Stinner, a python maintainer, shared another update that may be of
interest.

Joe

-------------------- Start of forwarded message --------------------
From: Victor Stinner <vstinner@python.org>
Date: Sun, 1 Oct 2023 12:36:46 +0200
Subject: Re: Python buildbots for FreeBSD
To: Joseph Mingrone <jrm@freebsd.org>

Hi,

Python now has 3 FreeBSD buildbot builders:

(1) FreeBSD 13.2-RELEASE 4 vCPU VM, 4GB RAM

owner: Zachary Ware <zachary.ware+bb@gmail.com>

(2) FreeBSD 14.0-BETA3, 8 GB vRAM, 8 vCores, amd64

owner: Kurt Jaeger <buildbot-fbsd@opsec.eu>

(3) FreeBSD 15.0-CURRENT, 8 GB vRAM, 8 vCores, amd64

owner:Kurt Jaeger <buildbot-fbsd@opsec.eu>

Two or three weeks ago, we also had a Cirrus CI FreeBSD job, but
sadly, we exceeded Cirrus CI "free" limits after 9 days, and Cirrus CI
cannot provide more CPU time to us.

So far, the 3 FreeBSD builders do they job: clone Python repository,
build the code, run the Python suite. So far, so good. So bugs
specific to FreeBSD have been discovered and there is a
work-in-progress to fix them. It seems like this proactive research
for FreeBSD-specific bugs is a new thing and I love it :-)

Example: KqueueSelector miscounts max events (*). Kqueue seems to be
specific to FreeBSD (I forgot if macOS uses it or not), and they bug
was not seen before while it likely exists since 2014, when the
selectors module was added to Python 3.4.
(*) https://github.com/python/cpython/issues/110038

Other examples of open issues which seem to be specific to FreeBSD, or
at least more likely to affect FreeBSD than other operating systems.

(*) asyncio: _UnixWritePipeTransport.write() may raise BrokenPipeError
on FreeBSD
https://github.com/python/cpython/issues/109757

(*) test_asyncio: test_start_tls_server_1() fails randomly on the
Cirrus CI FreeBSD job
https://github.com/python/cpython/issues/109051

What is needed right now are developers looking into these issues :-)

About the number of FreeBSD, I'm not sure if adding new builders is
"needed". It depends if you want to test a different CPU architecture,
a different FreeBSD configuration, etc.

Victor

-------------------- End of forwarded message --------------------

--=-=-=
Content-Type: application/pgp-signature; name="signature.asc"

-----BEGIN PGP SIGNATURE-----

iQKkBAEBCgCOFiEEVbCTpybDiFVxIrrVNqQMg7DW754FAmUdgdVfFIAAAAAALgAo
aXNzdWVyLWZwckBub3RhdGlvbnMub3BlbnBncC5maWZ0aGhvcnNlbWFuLm5ldDU1
QjA5M0E3MjZDMzg4NTU3MTIyQkFENTM2QTQwQzgzQjBENkVGOUUQHGpybUBmcmVl
YnNkLm9yZwAKCRA2pAyDsNbvnoo6EACtY33TD2QQjEkVcw3CYJuo8QbRkFmukvk9
Ef73jspmAhFVMWLDXq6lRGCdpKBzlB9cFL8RH6V/aWp4DeIUHQda2NDpmgpNfRBo
itlK/jteFnbGE2By7MUsdV7cRung0sMC6YOe8ykiiGKKW1hEKtXlt+WHrXMmz1l0
+iHRfbziL7TSLEC0uvPLCoGkmCZI+O0ylJXCV2aLMcY1aif0YImf5hQIV8eNe5Yn
BEGZpxn4v9RACYESgpGjlCrqs6+f3Uv1/PzI3QDeG/a5OH4gJDtCLQjQIoiJMUgY
NDXIpPZCjcIhqk0POEW9nJsmSUvn/lbUrzuUoH1wwlowS1UekcrsIJAKuuf0Qwh3
YRYIF3p6D1VNVF4+wsSl1XhWSw7BFf9IhNAeJVy2xJ8Rp3KR4u4rqE4ChvFskhK0
7xFrqiRsmkC2x2qENMaeVojlCWjFl5GBdfJFcmQiGF6gx29e8BCVrCufTwJJMveA
jiGhhMs85wvs6gf32zeHgEZmKcU2oqGkEUNr6ox1Fx3smiwsdnYbjHHUZIbt0FMT
WmKwXX8vQEEzZJ5S7L4G/W2iiY3tZeG6FDURWdYtnioif98WeQ3aR03Ey3bxgiFc
jHzZgY6DwqEAowFtCVrhWOdlJlKV4fAb51jPDLZ2znbDTw3epgBBxD8ml5ljL3aQ
adqcxudcdQ==
=RH/i
-----END PGP SIGNATURE-----
--=-=-=--



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?868r8ia04r.fsf>