Bug#766102: libcache-cache-perl: nondeterministic FTBFS: failure in t/6_test_size_aware_memory_cache.t

[Damyan Ivanov]

Control: forwarded -1 https://rt.cpan.org/Public/Bug/Display.html?id=88418

-=| Niko Tyni, 21.10.2014 00:06:23 +0300 |=-
> Package: libcache-cache-perl
> Version: 1.07-1
> 
> This package failed its autopkgtest suite recently on ci.debian.org:
>  http://ci.debian.net/data/packages/unstable/amd64/libc/libcache-cache-perl/20141020_123627.autopkgtest.log
> 
>  t/6_test_size_aware_memory_cache.t ......... 
>  1..48
>  [...]
>  ok 38
>  Couldn't limit size to 28 at /usr/share/perl5/Cache/CacheSizer.pm line 92.
>  ok 39
>  ok 40
>  ok 41
>  ok 42
>  ok 43
>  Couldn't limit size to 28 at /usr/share/perl5/Cache/CacheSizer.pm line 92.
>  ok 44
>  Use of uninitialized value $first_value in string eq at /usr/share/perl5/Cache/SizeAwareCacheTester.pm line 149.
>  not ok 45 # failed '$first_value eq $value'
>  ok 46
>  ok 47
>  ok 48
>  Failed 1/48 subtests 

Seems to be RT#88418. It also has a patch, but I did not test it.

> It's hard to reproduce, but I managed to trigger it on my own sid/amd64
> system by running the test in a loop:
> 
>  while prove t/6_test_size_aware_memory_cache.t; do : ; done
> 
> which eventually (after 5-10 minutes) failed once with the same symptoms.
> I also got it to trigger similarly with 'prove -b' in a just-built source
> tree so it's not specific to the autopkgtest setup and can cause sporadic
> build failures.
> 
> Filing at severity:normal because of the difficulty in triggering it.
> Probably load related? I expect ci.debian.net is quite busy.

The RT ticket seems to suggest that an overloaded environment would be 
helping triggering the issue indeed.