t.lock2 issues

[David Relson]

At 07:59 AM 2/1/03, Matthias Andree wrote:

>On Fri, 31 Jan 2003, David Relson wrote:
>
>
>OTOH, the changes aren't dangerous, so let's use the patch, seed the
>stuff with the microseconds as you suggest and change the suggested
>cast to (float) to (double) because we use double all over the place and
>we don't want to sacrifice precision (in terms of the mantissa) -- float
>has 23 bits usually (losing 11 bits), double has 52 bits.

O.K.  I'll do it.  If we #define the MIN/MAX as floating point values, we 
can let the compiler do the conversion.  For readibility, I'm going to 
isolate the whole sleep calculation in its own function.