|
|
Boost Testing : |
From: Markus Schöpflin (markus.schoepflin_at_[hidden])
Date: 2005-05-17 03:07:25
Hartmut Kaiser wrote:
>
> Markus Schöpflin wrote:
>
>>Seems I spoke too soon. :-/ There are still some failures
>>when running the regression tests. When looking at the errors
>>from gcc-3.4.3 (see http://tinyurl.com/a7rsx), it strongly
>>smells like some 64 bit issue. Maybe using a long instead of
>>int32_t or something like that. Or maybe the tests are flawed?
>
> Yes, the tests are flawed. I'm currently not sure what to do about this,
> though. Will have to think a bit.
>
> WRT the current result on the True64 platform: there is something really
> broken and I have no clue, what happened, especially since it already worked
> there (perhaps except the 64Bit issues above).
You mean Tru64/cxx as opposed to Tru64/g++? I wouldn't be too sure that it
really worked, because of the error that was present in process_jam_log
which lead to erroneous success reports. I have been digging around a
little but so far I didn't find anything promising.
BTW, is wave using static local variables in templates which get
instantiated implicitely? If yes, that could be a reason because the
compiler will generate multiple instances of such a variable, one for each
instantiation.
Markus