Re: Arithmetic type emulator helper class for testcases.

["Chris Fairles" <chris dot fairles at gmail dot com>]

On Sat, Jul 12, 2008 at 1:27 PM, Paolo Carlini <paolo.carlini@oracle.com> wrote:
> Hi,
>
>> I just wanted some feedback based on new information I've
>> gathered regarding the clocks and POSIX systems.
>
> Always remember, however, that we cannot assume POSIX, we can only in general assume C89, anything beyond it requires configure tests.
>
>                                                                  From what I gather,
>> gettimeofday() *must* be available to all posix systems but
>> only provides mircosecond resolution whereas clock_gettime (a
>> POSIX.1b real-time extension) gives nanosecond resolution but is only
>> available if CLOCK_REALTIME is defined, and the monotonic version is
>> available only if CLOCK_MONOTONIC is defined.
>
> I think you need 2 tests (TRY_COMPILE is enough): the first one for the availability of clock_gettime in <time.h>; then another for unistd.h defining _POSIX_MONOTONIC_CLOCK (you have lots of examples in acinclude.m4 about checking for the availability of a function, checking for the preliminary availability of unistd.h, etc). The constants you mentioned above are just possible values of the first argument of clock_gettime, not something to be used at configure time. Please take time to experiment a bit, there are always various, stricter and less strict, ways to perform such tests. I suggest using as a reference the various entries (in a particular, unistd.h, and clock_gettime) of http://www.unix.org/single_unix_specification/
>
> Paolo.
>
>

Ok, I attached a patch (for review purposes) of what I currently have.
Is the first check at least ok? The second check I don't like. The
POSIX std says _POSIX_MONOTONIC_CLOCK "shall be defined to be -1, 0,
or 200112L" (if available) and also "the value of this symbol reported
by sysconf() shall either be -1 or 200112L". On my system,
_POSIX_MONOTONIC_CLOCK is 0 and sysconf returns 176. Since its greater
than 0 (whatever that means) it leads me to believe that I'm not
interpreting the std correctly.

Maybe I should just test for _POSIX_MONOTONIC_CLOCK > 0 for success so
if there's ambiguity because its defined to be 0, its considered not
supported.

help!? :)

Chris

Attachment: acinclude_rtclock_patch.txt
Description: Text document