On Thu, 18 Dec 2008, Isidor Zeuner wrote:

> I'd be glad to receive comments if someone finds it useful or interesting.

I find it both useful and interesting! I like it and I want to see this
applied so that we truly get sub-second resolution on most operations even
with synchronous name resolves!

In addition to the not C89 compliance remark, I'd like to see the patch fix so
that there's less #ifdefs. Possibly this can be accomplished by introducing a
set_alarm() function and a restore_alarm(), that would use the necessary and
available functionality. Adding this rather complicated functionality - as at
least the switch-back logic is quite complex - inline in the funtion is making
it hard to read!

-- 
  / daniel.haxx.se