Date: Sat, 22 Dec 2012 07:34:36 +1100 From: Kubilay Kocak <koobs.freebsd@gmail.com> To: mva@freebsd.org Cc: "python@freebsd.org >> \"python@freebsd.org\"" <python@freebsd.org> Subject: Re: lang/python33's make test target creates artifacts and long build times Message-ID: <50D4C7DC.7020703@gmail.com> In-Reply-To: <20121221085210.GC1993@medusa.sysfault.org> References: <20121221085210.GC1993@medusa.sysfault.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On 21/12/2012 7:52 PM, Marcus von Appen wrote:
> Hi,
>
> today I noticed some issues with the tinderbox runs of lang/python33
>
> First of all, the test: target introduced causes artifacts within
> ${PREFIX}, effectively polluting the user environment. Is this really
> intended?
>
> In my opinion a port, that executes the test target, should do this either
> in a separate directory/environment to avoid cluttering the user's
> environment with garbage (well, it's Python's fault here in my opinion
> to use a test/data directory for the test results).
>
> However, a user will most likely not execute a test target within a typical
> installation procedure, tinderbox however does, causing two issues:
>
> - around 15 test files are created in %%PYTHON_SITELIBDIR%%/test/data
> - the python test cases randomly crash and require about 45 minutes to
> run, whereas they take about 5 minutes to run outside of a tinderbox
> and never crash[1]
>
> The usefulness of the test target hence is (at the moment) zero,
> contradicting its purpose.
>
> I would strongly advocate to back it out asap.
> If there are no other opinions on that, I'll do so on the next Sunday.
>
> [1] on my box of course :-)
Yup, was aware of the leftovers but obviously didn't get around to
resolving it.
+1 on removing the the test target until i/we can figure out a way to
avoid the spurious file creation (or cleanup after)
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?50D4C7DC.7020703>