On Tue Jul 29 16:12:12 2014, coke wrote:
> last 2 days we've started seeing thousands of failures on the daily
> rakudo.parrot run. Doing a git bisect, looks like the commit that
> started the failures is:
>
> $ git bisect good
> 9515bb7636d2352a68bf770d80964f5907526fbc is the first bad commit
> commit 9515bb7636d2352a68bf770d80964f5907526fbc
> Author: jnthn <jn...@jnthn.net>
> Date: Sun Jul 27 22:22:59 2014 +0200
>
> Tweak some protos for better Junction failover.
>
> We can make the proto itself be (Any, Any), meaning it fails to bind.
> However, this also means that the multi-dispatcher never gets chance
> to sneak an entry into the cache for the Junction case, so we have to
> go through the slow-path binder to fail over every time, rather than
> cheaply identifying we should hit the Junction dispatcher.
>
> ... whether or not my git bisect is right, here's a golf of one of the
> currently aborting spectests:
>
> 19:10 < [Coke]> r: use Test; is(:10('01110') , 0d1110, ":10('01110')
> is default
> decimal");
> 19:10 <+camelia> rakudo-{jvm,moar} c00999: OUTPUT«ok 1 - :10('01110')
> is default
> decimal»
> 19:10 <+camelia> ..rakudo-parrot c00999: OUTPUT«(signal )»
The commit that caused this was re-worked to avoid the issue. Closing ticket.
--
Will "Coke" Coleda
