D issues are now tracked on GitHub. This Bugzilla instance remains as a read-only archive.
Issue 18103 - test17868 failing often but not every time
Summary: test17868 failing often but not every time
Status: RESOLVED FIXED
Alias: None
Product: D
Classification: Unclassified
Component: dmd (show other issues)
Version: D2
Hardware: All Mac OS X
: P1 critical
Assignee: No Owner
URL:
Keywords: TestSuite
Depends on:
Blocks:
 
Reported: 2017-12-18 22:55 UTC by Brad Roberts
Modified: 2020-06-19 06:47 UTC (History)
4 users (show)

See Also:


Attachments

Note You need to log in before you can comment on or make changes to this issue.
Description Brad Roberts 2017-12-18 22:55:47 UTC
Not sure when it started exactly, but for a while test17868 in the dmd test suite has been failing.  Seems to be limited to osx,

https://auto-tester.puremagic.com/platform-history.ghtml?projectid=1&os=Darwin_64_64

https://auto-tester.puremagic.com/platform-history.ghtml?projectid=1&os=Darwin_64_32

It's not tied to a single host, all three hosts seem to have a mix of passing and failing.
Comment 1 Steven Schveighoffer 2017-12-19 15:19:57 UTC
Looks like its a race condition.

The failure line says this:

diff: generated/runnable/test17868.d.out.2: No such file or directory

Note that the postscript does something with the output, copies it to a ".out.2" and then diffs it against the expected output.

But I see no "release" or "debug" in that path. The appropriate line is this:

diff --strip-trailing-cr runnable/extra-files/test17868.d.out ${RESULTS_DIR}/runnable/test17868.d.out.2

So clearly, RESULTS_DIR just contains "generated". That's not enough to avoid races. I actually see a lot of cases in the postscript files that are subject to possible races.

CCing Martin, as he seems to be the one that set up these tests.
Comment 2 moonlightsentinel 2020-06-19 06:47:03 UTC
That test doesn't use a POST_SCRIPT anymore.