[DRE-maint] Bug#836536: ruby-concurrent: FTBFS too much often (failed examples)

[Santiago Vila]

On Mon, 12 Sep 2016, Antonio Terceiro wrote:

> On Mon, Sep 12, 2016 at 11:12:26AM +0530, Balasankar C wrote:
> > On Sat, 3 Sep 2016 21:49:13 +0200 (CEST) Santiago Vila <sanvila at unex.es> wrote:
> > >   BTW: I think the "Randomized with seed xxx" above is a bad idea.
> > > Maintainers often ask bug reporters for "a way to reproduce the error".
> > > But if the build is based on random variables, this becomes an
> > > impossible task. Please consider modifying the build system so that
> > > those random things are always the same.
> > 
> > I believe I have fixed this in the latest upload, in which I specified the
> > order to run the tests. I tried building it 5 times in a cowbuilder environment
> > and didn't hit any failures. Please reopen the bug if this problem persists.
> 
> I think that in general 5 times is probably not enough. the last package
> I fixed that failed randomly would fail 6% of the time (i.e. ~3 times
> out of 50 attempts in my tests).

Indeed, for such a low probability it's better to try a lot of times
to be sure. For this package, I have just built it 200 times and it
failed none, so I believe that yes, it's fixed.

[ Note: I have unarchived this bug to change the Subject, I never
  liked the "too much often" wording. Either it fails always,
  or it fails randomly. How often it fails should not really matter ]

Thanks.