This is the mail archive of the libstdc++@gcc.gnu.org mailing list for the libstdc++ project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Arithmetic type emulator helper class for testcases.


On Tue, Jul 15, 2008 at 6:13 PM, Paolo Carlini <paolo.carlini@oracle.com> wrote:
> Hi,
>
>> > 2- Double check that the new testcases are -Wall clean (make check CXXFLAGS="-O2 -Wall"
>>
>> Added __attribute__((unused)).
>
> But in general try to keep those to a minimum. Very often there is a way...
>

Can do. I thought this attribute was ok because its used all over the
place. Easy enough to suppress the warning by other means.

>> No real reason why it can't be dg-do run.
>
> Then just remove it, dg-do run is the default. Also, double check the other files for the dg-do * *first* in the file issue, I spotted at least a wrong one.
>

Done and done.

>  I just did link
>> because I wasn't making any run-time checks. But, I'm probably going to
>> revisit the test cases to harden them up once the clock_gettime/-lrt
>> nonsense
>
> By the way, if you have something decently clean to suggest... help welcome! You said there is something on the web about that, about dealing with the different cases? A problem I can see is feeding back that info to Makefile.am. But really, any suggestion is welcome, I think we delayed seriously attacking this issue (adding libs on a case-by-case basis) for far too much time.
>
> Thanks,
> Paolo.
>
>

I'll see what I can do (re: 3rd party lib configuration) while looking
into librt. I'll follow up in a new post.

Chris

Attachment: chrono_et_al_patch.txt
Description: Text document


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]