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: Replace fork with try_fork in testsuite?


Mark Mitchell wrote:

> Paolo Carlini wrote:
>
>> Mark Mitchell wrote:
>>
>>> If you want me to add HAVE_{MKFIFO,FORK} let me know, and I'll fix
>>> try_mkfifo while I'm at it.
>>
>> Indeed, this is what I was thinking...
>
> OK.

Hummm, Mark, I'm sorry, but now an important point comes to my mind,
that we briefly discussed with Benjamin a few months ago.

Don't you think that when such facilities are not available, would be a
lot better not running such tests at all? Otherwise, in the final
DejeGNU report you see always the same number (think gcc-testresults)
and cannot figure out immediately which tests that machine is *actually*
running.

This is much more general than FORK and FIFO, indeed. Extends to other
classes of tests.

What do you think, *in principle*?

Paolo.


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