> (1) tick + tock return elapsed real time as well as estimated run time...
> (2) tick + tock are not tied to output labels...
Sorry, I didn't mean the existing time function which is indeed
limited as you say.
Let's call it fulltime() or have two functions realtime() and runtime().
> ...I guess it could be argued that exposing get-internal-real-time is less ugly than tick + tock.
> get-internal-real-time and get-internal-run-time return integers.
> Just carrying them over into Maxima preserves every bit,...
I would scale them to (b)floats in seconds, removing the
implementation dependency. Runtime zero would start at process start,
presumably, and realtime zero would be at some convenient epoch, e.g.
1 January 2000. Float representation would be chosen to have a
resolution of at most 10 nS, which is much more accurate than
necessary given overhead and constraints on reproducibility. If the
accuracy needs to be improved in the future, the interface does not
need to change. Rationals are cute for this, but I don't think have
any practical advantages. Anyone who wants more accurate timings will
probably have to do low-level coding anyway....
-s