[rfc][cedtea-web] another reproducers stabilization

[Andrew Azores]

On 01/16/2014 09:50 AM, Jiri Vanek wrote:
> On 01/16/2014 03:45 PM, Andrew Azores wrote:
>> On 01/16/2014 08:58 AM, Jiri Vanek wrote:
>>> Thsi is removing more xception checks from itw. By this patch 
>>> another 300 reproducers will become
>>> stably green. Unluckily, the revert of
>>> http://mail.openjdk.java.net/pipermail/distro-pkg-dev/2014-January/025764.html 
>>> is stil needed, and
>>> intersection of fixed tests is zero :(
>>>
>>> J.
>>
>> Why are these checks now invalid? Are the exceptions no longer 
>> printed for some reason, or are they
>> going to stdout rather than stderr? Or are they now catching false 
>> negatives ie exceptions are
>
> yes
>
>> occurring that do not actually signify test failures?
>>
>> Thanks,
>>
>
>
> They were ensuring  that *no* *general* exception occurred. Unluckily 
> in many test runs some completely unrelated, and absolutely harmless 
> exceptions  are printed.  The torturing of browsers/java by repeated 
> kill -9 do this.
>
> So Although at first the tests were "pessimists" == everything have to 
> be ok. Now they are "optimistic" == just the core of reproducer passed.
>
> Although I'm fan of "pessimistic"  tests. In this case it caused the 
> great instability of engine - random failures each run.
>
> J.
>
>

Okay, that's what I thought was going on. Seems fine to me then. I'd 
prefer pessimistic testing, too, but it's not worth it if it incurs 
random failures and instability.

Thanks,

-- 
Andrew A