[PATCH] gettimeofday time travels

Philip Aston paston@bea.com
Tue Apr 16 01:43:00 GMT 2002

cgf writes:
 > On Mon, Apr 15, 2002 at 09:58:48PM +0100, Philip Aston wrote:
 > >How about the attached quick and dirty fix?
 > I'm sorry but I don't think a quick and dirty fix is justified if
 > there are other alternatives. I haven't seen any other alternatives
 > discussed yet.

Understood and appreciated. Here's some justificiation/discussion for
my fix.

I'm no W32 expert but:

1. Suspend and resume changes the counters in a seemingly
   unpredictable way. (I'd have guessed that the counters would have
   been "suspended" and hence the reported time after a suspend would
   be in the past). I wouldn't want to patch into suspend notification
   API's to implement this.

2. There are other situations in which the base time could
   be invalidated (NTP update, timezone change, user clock change).
   How can/should we detect and react to these events?

3. There is a chance that QueryPerformance* _should_ take
   suspend/resume into account and that my hardware is at fault. Any
   volunteers to run my test case on their machines?

- Phil

Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

More information about the Cygwin mailing list