|
|
Boost Testing : |
From: Eric Niebler (eric_at_[hidden])
Date: 2007-11-15 11:18:48
Beman Dawes wrote:
> Rene Rivera wrote:
>> Eric Niebler wrote:
>>> Rene Rivera wrote:
>>>> Eric Niebler wrote:
>>>>> The xpressive failure here <http://tinyurl.com/2srknl> is due to an
>>>>> expired timeout. I think 300 seconds is too low. Can the timeout be raised?
>>>> Do you mean can the default be increased? Or the timeout in that
>>>> particular machine?
>>> Well, I meant the timeout on that machine, but now that you mention it,
>>> it may be a good idea to raise the default, too. Maybe 10 or even 20 min?
>> Well, it seems to me that having file take more than five minutes to
>> compile is rather bad. And hence may be a problem in the file. So I'm
>> reluctant to increase the limit too much. And in this particular case I
>> don't think the jrp machine is a slow one. So perhaps some investigation
>> as to why this particular test is taking a long time on that machine is
>> warranted?
>
> FWIW, I share Rene's concerns.
Jeez, sometimes it take bjam 5 minutes just to figure out what to build!
;-) Like I said in a previous message, there is nothing wrong with the
file. It's a big, complicated meta-program that take 1min 30sec to
compile on my laptop. Clearly this test machine is struggling with it.
It just needs more time.
The timeout is to catch runaway tests. We expect that to be a rare
occurrence, so a long timeout should be ok. A short timeout is just
going to cut off tests that are working and result in spurious failures.
-- Eric Niebler Boost Consulting www.boost-consulting.com