Re: PATCH: my "SECONDS can be floating point" tweaks

[Wayne Davison <wayned@xxxxxxxxxxxxxxxxxxxxx>]

On Thu, Oct 31, 2002 at 10:34:45AM +0000, Peter Stephenson wrote:
> Looks OK, although I still vaguely think a struct timeval is more
> appropriate for the raw values (which doesn't fit inside struct param).

That is certainly an option, and it would be easy to tweak my code to
implement this if it is desired (after adding a "struct timeval" to the
param.u union).

In my limited testing the sub-second value was restored exactly, but I
can imagine that certain fractional representations would cause the
sub-second value to get tweaked by a small amount the first time it was
saved and restored (and remain constant after that).  I haven't checked
yet to know for sure how big this tweak could get, but I'll look at it
more closely soon.  (My gut says that the values will be close enough
that the change will not be noticed.)

I've gone ahead and committed my previous patch.

..wayne..