Re: [PATCH, v3] Potential solution to librt issue.

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

On Thu, Jul 24, 2008 at 8:37 AM, Paolo Carlini <paolo.carlini@oracle.com> wrote:
>> The only issue I see with removing clock_gettime is that I'm
>> not sure if posix systems have other options for a monotoic clock.
>
> As far as I know, they don't. That would be a huge problem, then, as far as I can see. And, I'm sorry, talking generically about "no-no" doesn't make sense here: these timers are C++0x features, already in the working draft, either we maintain that the facility is not implementable on linux (and we report about that to the ISO C++ Committee) or we find a way to do that. Sure, we can delay dealing with the issue, but how much, 1 year, two? Can I ask the constructive help of the knowledgeable people here?
>
> Anyway, getting to the technical point, the use of librt proper is an implementation detail, the user of <chrono> cannot know about it, neither he can know that using a timer in <chrono> requires for him to explicitly link libpthread. Is that libpthread thing confirmed? In that case, what we have now is already broken and we are going to revert it. But please, help properly solving the issue anyway.
>
> Paolo.
>
>

I stumbled upon this C 200x proposal:
http://www.twinsun.com/tz/timeapi.html
(from http://www.cl.cam.ac.uk/~mgk25/time)

I haven't been following any developments on the C 200x front and a
quick search didn't find an accepted version of this proposal so I
don't recommend waiting for the C-folk to do the dirty work for us ;)

I suppose there's no way to use clock() (i.e. cpu-time) for the chrono clocks?

Chris