fpco/stackage-curator

Print why something failed

Closed this issue · 1 comments

We continue experimenting with using Stackage for regression-testing of GHC (see recent blog for more details: https://www.tweag.io/posts/2018-04-17-stackage-head-is-live.html). And we detect stuff like this:

streaming-commons:
  at nightly-2018-04-13-4d30bc869dba8480fad2487911e828aab6ffd53c.csv
    build succeeded, 1 test suites passed, 0 test suites failed
  at nightly-2018-04-13-5d76846405240c051b00cddcda9d8d02c880968e.csv
    build succeeded, 0 test suites passed, 1 test suites failed

Which corresponds to log containing this sort of thing (at GHC commit 5d76846405240c051b00cddcda9d8d02c880968e):

Test run streaming-commons-0.2.0.0 (test) (pending: 130, failures: 818)
streaming-commons: Process exited with ExitFailure 1: dist/build/test/test (pending: 81, failures: 828)

But we can't reproduce it afterwards. The test passes. If stackage-curator could print why something failed (i.e. attach output of a process before it died or save it somewhere for later inspection) that would help a lot. Do you think this is possible to add, would you be open to a PR implementing something like this?

Thanks!

Ah so it creates log files. I was blind. Closing.