Re: threading / timers / etc

[Andy Spencer <andy753421 gmail com>]

On 2010-10-22 17:06, Ryan Lortie wrote:                                
> - We will add GTimeSpec which is GTimeVal using nanoseconds instead
>   of microseconds (same story with struct timeval vs. struct
>   timespec).  The librt interfaces provide this level of accuracy so
>   it would be a shame to needlessly discard it.

So now we have seconds, microseconds, and nanoseconds? I'm sure this is
all for compatibility with librt, but maybe we could cut to the chase,
use a guint64, and represent it as attoseconds? If we do the same for
seconds (gint64) we could conveniently represent any currently
measurable time interval in the known universe.. and then avoid adding
something involving picoseconds in a few more years.

Bonus points if it makes consistent use of leap seconds.