🚀 go-pugleaf

   >Date: 29 Jan 90 23:02:30 GMT
   >From: nicholls%bcsaic%ssc-vax@beaver.cs.washington.edu  (Bill Nicholls)
   >Subject: More on the real time clock.
   >
   >Remember the real time clock problems a few weeks back?
   >
   >Here's a new one.  I'm trying to get Rel6 running on my 
   >explorer II.  All is well except that every time I reboot
   >the machine, initialize-name-service takes whatever time
   >is in the real time clock, subtracts two hours from it,
   >then puts it back again.  We are two hours from Texas.
   >
   >I can prevent this behavior by having a nil
   >'primary-time-servers' attribute in my site namespace object.
   >But, that is not the preferred situation.
   >
   >Any ideas??
   >Bill

This looks like one of the favourite bugs in the Explorer system - we've
patched it ever since rel 3.2.

The BOOT namespace is created with a site object with :TIMEZONE 6,
hardwired, and when your real namespace comes up with some other
:TIMEZONE, you get your time shifted by the difference.  A work-around
for this is to patch NAME:INITIALIZE-DISTRIBUTION-NAMESPACE to use a
variable instead of "inline-coded" 6.  Then set this variable, do a
disk-save and it should work.

This really is a work-around, since it implies that you have to set this
silly variable to whatever timezone you're in, defeating the purpose of
the namespace.  There are a few real problems with getting the time.
You shouldn't ask yourself, and you should make sure you get the right
timezone.  We also wrote a new function that applies NET:HOST-TIME to
NET:TIME-SERVER-HOSTS *except* the local host, and set
NET:*NETWORK-TIME-FUNCTION* to this new function.

--Bjorn Victor		       Victor@John-Jameson.DoCS.UU.SE
Dept. of Computer Systems       or
Uppsala University, SWEDEN     Victor%John-Jameson.DoCS.UU.SE@uunet.UU.NET