On Mon, 10 Oct 2016 01:51:37 +0200, Emmanuel Bourg wrote:
Le 10/10/2016 à 01:40, Gilles a écrit :
For months, Travis consistently did not work (i.e. even when
that non-deterministic passed, which it does 99% of the time
for truly uniform RNG) even though local and Jenkins builds
went fine.
The problem reported as MATH-1387 was about what was wrong with
the Travis configuration. [Thanks to Rob for fixing it.]
The bug isn't titled "The Travis configuration is broken", it's
titled
"Travis does not work",
My point, exactly: it has been discussed (ML) that something
was broken with Travis (be it configuration, as it turned out
or something in its environment) since the build was OK otherwise.
this means the resolution may include
configuration fixes *and* test stabilization.
What you call "test stabilization" did strictly _nothing_
to make Travis work.
Rob even pasted the log of
the test failure in the bug so the intent of this commit is clear.
The commits were fine (separate issues in separate commits).
Only the message was not correct as far as the "RetryRunner"
change is concerned.
Not a big deal, but confusing.
Enough talk about this as far as I'm concerned.
Gilles
Emmanuel Bourg
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
