At Mozilla startup, NetscapeDebug says it will expire October, 1994. Not a real problem, but disconcerting. If this is based on some system call, is all time related code in Mozilla for MacOS wrong?

Reassinging time bomb stuff to cyeh.

reassigning to sdagley, king of the time routines. Have verified that timebomb number generation is correct and working. Steve, the timebomb date displayed is correct with regards to the Month and Day, only the year is incorrect and displays 1994. The timebomb works properly, so it's only the displayed expiration time that is incorrect.

Happy happy, joy joy, more time code weirdness to figure out (as if I didn't kill enough in 4.x :-) ). Turning into a platform parity bug since I'd bet dollars to donuts it's a Mac specific bug in either the NSPR or JavaScript code. Adding WTC to the cc: list in case the NSPR team has a time routine test applet we can build on the Mac to test against.

is this mac only wierdness for time, or is this a general xp timebomb problem?

What is the correct year? Steve, would you like to schedule a one-hour debug session together? There is a simple test case timemac.c and a complicated test case timetest.c in mozilla/nsprpub/pr/tests. However, we don't have the CodeWarrior project files for the test cases.

A debug session sounds fine and I'm pretty flexible on time this week except for Thursday when I'm planning on being at the MacWorld Expo in San Francisco.

So far we've discovered that the file mactime.c exists in 3 places in the mozilla source tree and the one in the path mozilla/lib/mac/Misc contained a bad implementation of the GMTDelta() routine (was using the Mac Toolbox call BitTst to test the wrong bit). Still need to review the NSPR 2.0 code to verify the multitude of Mac time code fixes from NSPR 1.0 code were picked up.

NSPR 2.0 time test app build now available on Mac and indicates no problems. We still have an issue with caching the machine location at startup to address a problem on older 68K Macs (this is the problem that Netscape Defrost was created for) which we probably don't need/want on PPC machines but that isn't an NSPR problem per se.

sdagley, you have the test app. Can I mark Verified now?

That would be a yes. Marking verified.

Inserting Milestone info.