[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]
January 19, 2038
- From: Dan Morrill <morrildl nycap rr com>
- To: axp-list redhat com
- Subject: January 19, 2038
- Date: Thu, 28 Jan 1999 11:28:04 -0500 (EST)
I have read that the "UNIX" clock's mechanism of counting the seconds
since January 1, 1970 will exceed the range of the 32 bit field used to
store it on January 19, 2038 at 3:14:07.
Now, that can be "fixed" by updating the time field to 64 bits, but that
of course causes problems unless all the C libs, data, and applications
that query the date are also updated.
I was thinking that perhaps I'd set my Alpha's clock to Jan. 1, 2038
3:14:00 and watch what happened 10 seconds later, but then I decided that
that's probably playing with fire. (I determined, when RH 5.1 set my date
to 2018, and once to 2035 that Linux/Alpha is Y2K compliant, but that's
not the same as 2038.)
So, instead I'm asking a few questions of you gurus out there:
* does Linux/Alpha already use a 64 bit time field? (I'm guessing not,
as that would probably introduce incompatibilities)
* how big is the hardware clock field in Alpha mainboards (in my case,
SX164), and what would happen to my hardware if I were to set the date
to something untoward?
* how much of a nuisance is this going to be (i.e. updating the kernel
and glibc) come 2038?
I'm not especially worried, partly because it's still 40 years away, but
mostly because this is precisely the sort of thing open source is good at
flushing out, but I thought I'd ask anyway.
Dan Morrill
Computer Scientist, Physicist
[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]
[]