times off by one hour

[Andrew Tridgell]

> The problem I described with the print job times being a day off still
> exists (and is apparently completely independant from the timezone
> setting of the NT box).

I've now had a chance to look at this (and have reproduced it) and it
appears to be a NT bug.

The relevant date field is set in ipc.c/fill_printjob_info(). It is a
32 bit time since 1970 field and thus is not subject to off-by-one
errors in the day. I have also tracked the field all the way back to
the initial time() call and no changes are made to the OS supplied
time. 

I think that NT4 displays the date incorrectly when parsing the ipc
response. We could add a ugly hack to add one day in that function but
I don't think we should.

When NT4 talks to a NT server I presume it uses different calls
(perhaps pipe calls?) which would explain NT server "getting it
right".

	Cheers, Tridge