|
|
Boost Testing : |
From: Eric Niebler (eric_at_[hidden])
Date: 2007-06-03 17:56:15
Martin Wille wrote:
> Eric Niebler wrote:
>> Rene Rivera wrote:
>>> Eric Niebler wrote:
>>>> Douglas Gregor wrote:
>>>>> |xpressive|
>>>>> test7: gcc-3.3.6_linux
>>>> Huh, this test has been passing for ages and the code hasn't changed
>>>> recently. Spurious failure?
>>> Probably spurious. Marting has some random memory errors on the RAM of
>>> those machines, as in a single bit gets flipped with a low probability.
>> Then can I humbly suggest that Martin refrain from running any further
>> tests on this machine until the memory problem is fixed? Wrong test
>> results are worse then no results at all IMO, and will only slow us down.
>
> That would mean I won't run any further tests on this machine at all, as
> I'm not going to waste any money for a 6 years old box. I'm not sure
> this is desired.
That's unfortunate, but I still believe that no results are better than
incorrect results. Failures on known-bad hardware will be ignored. What
good is a failure report that is ignored? It's nothing but a
distraction. We need reliable test results.
Your contribution is valuable. RAM is cheap. Shall we take up a
collection? :-)
-- Eric Niebler Boost Consulting www.boost-consulting.com