On 21/06/16 02:16, Frank Ch. Eigler wrote:
kenj wrote:
[...]
666 remains a royal pain in the backside
[...]
Several of the pending patches on pcpfans.git fche/pmfuntools (nee
fche/multithread) were written 5-7 weeks ago to improve this.
Thanks for the pointer Frank. These commits have been merged now (based
on a quick inspection of the commit history in the two trees) and on the
one machine I've tried, qa/666 has passed 3/3 attempts ... so that's
very encouraging.
As one who spends more time debugging other people's qa tests than
anyone else (which is not related to 666 but is a more general
statement), it is however a bit perplexing that 666 runs for more than
five minutes, especially as the rest of the QA suite has running times
where 90% of the tests complete in less than 20 seconds and 98% of them
are done inside 100 seconds.
As a general rule, fault isolation and debugging would be assisted by
tests that have narrow focus, even if this means multiple tests to
provide coverage for a complex piece of functionality ... I am not
suggesting refactoring 666, just pointing out my preference for others
developing new tests.
Back to my other 100+ failures ...
Cheers, Ken.
|