develooper Front page | perl.perl5.porters | Postings from August 2019

[perl #133771] ext/File-Find/t/*.t: Continued intermittent failuresduring parallel testing

From:
James E Keenan via RT
Date:
August 7, 2019 13:42
Subject:
[perl #133771] ext/File-Find/t/*.t: Continued intermittent failuresduring parallel testing
Message ID:
rt-4.0.24-2903-1565185317-499.133771-15-0@perl.org
On Wed, 13 Feb 2019 23:38:55 GMT, tonyc wrote:
> On Mon, 28 Jan 2019 14:52:18 -0800, jkeenan wrote:
> > On Wed, 23 Jan 2019 20:24:27 GMT, jkeenan wrote:
> > > On Mon, 21 Jan 2019 22:52:57 GMT, tonyc wrote:
> > > > On Mon, 21 Jan 2019 12:09:17 -0800, jkeenan wrote:
> > > > > On Fri, 18 Jan 2019 02:55:24 GMT, tonyc wrote:
> > > > > > On Thu, Jan 17, 2019 at 05:02:29AM -0800, James E Keenan via RT
> > > > > > wrote:
> > > > > > > Let me pose this question:  Would we be better off using
> > > > > > > File::Spec-
> > > > > > > > tmpdir() to create the top-level of the testing
> > > > > > > > directories,
> > > > > > > > then
> > > > > > > creating for_find/ underneath that directory (in most cases,
> > > > > > > /tmp)?
> > > > > >
> > > > > > I don't think it will make a difference over just searching
> > > > > > within
> > > > > > t/for_find.
> > > > > >
> > > > > > Tony
> > > > >
> > > > > Please review patch 0001-find.t-Use-temporary-testing-directory-
> > > > > for-
> > > > > all-block.201901211500.patch, attached.
> > > >
> > > > That looks better.
> > > >
> > > > Tony
> > >
> > > Merged to blead in commit 0abd1628824e7e5b7d6df370fcf5c143bf2dab8d.
> > > Will monitor ticket for several days before closing.
> > >
> > > Thank you very much.
> > 
> > Unfortunately, we don't have the problem licked.  See this smoke-test
> > report:
> > 
> > http://perl5.test-smoke.org/report/79324
> > 
> > #####
> > Test failures:
> > ~~ ../ext/File-Find/t/taint.t ..................................
> > FAILED Non-zero exit status: 2 Bad plan.  You planned 45 tests but ran
> > 1.
> > [stdio] -Duseithreads -Doptimize="-O2 -pipe -fstack-protector -fno-
> > strict-aliasing" -Dcc="gcc" -Duselongdouble DEBUGGING
> > #####
> > 
> > Here it is taint.t that is experiencing failures; the patch applied
> > dealt with find.t.
> > 
> > I doubt that the particular ./Configure options had anything to do
> > with the failure.  The smokecurrent.log didn't have any more
> > information than is visible at the URL cited above.
> > 
> > Needs further investigation.
> > 
> > Thank you very much.
> 
> Has this happened again since then?
> 
> It might be due to something unrelated.
> 
> Tony

See: http://perl5.test-smoke.org/submatrix?test=../ext/File-Find/t/taint.t

The failures from FreeBSD-11 are on the host where I smoke-test blead.  In my most frequently used configuration on that host, perl is configured, built and tested 36 times.  I get 1 or 2 instances of the failure in taint.t every 2 or 3 smoke-test runs.

I have stared at the relevant code off and on for months, but since I have no reliable way of generating the race condition, I can't test any insights that I might have.

I think we would be better off (where "better off" means getting us to the point where we can close this ticket) simply restoring ext/File-Find/t to the list of directories which are tested in series (currently at line 192 of t/harness).

Please review the patch attached.

Thank you very much.

-- 
James E Keenan (jkeenan@cpan.org)

---
via perlbug:  queue: perl5 status: open
https://rt.perl.org/Ticket/Display.html?id=133771



nntp.perl.org: Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About