Re: ProDOS year ranges (was: No-Slot Clock)

[dempson@actrix.gen.nz (David Empson)]

Paul R. Santa-Maria <paulrsm@ameritech.net> wrote:

> I have read that tech note, and I have always thought that Apple chose the
> wrong range just to be compatible with the Macintosh.

The only Macintosh compatibility issue I can think of is that the oldest
Mac models' 32-bit clocks will run out around 2040.  Later models (even
as far back as the LC) use a base date of 1956 instead of 1904, so they
will last to 2092.  The Mac also has 64-bit date calculation routines,
which support dates at least as far as 20000 BC and 20000 AD.

The main reason for the 1940/2040 range is that this is how the IIgs
clock routines work, for the same reason as the older Mac models: the
IIgs clock is a 32-bit counter which returns the number of seconds since
1st January 1904, so it runs out in 2040.

> The system clock is suposed to to tell you what the time is right NOW.

But the system clock is only part of the problem.  The main problem, as
I see it, is time stamps on files, which are usually derived from the
system clock (but not if a file is copied from elsewhere).

If we have two standards for how 2000 and later are represented, then
all implementations of the ProDOS file system (and all applications
which display file dates) will have to support both representations.

> That is just the system clock.  How an application stores and handles dates
> is an entirely separate issue.

Internally, yes.  They still need to deal with exporting and importing
dates (particularly file creation and modification dates) in a standard
form.

-- 
David Empson
dempson@actrix.gen.nz
Snail mail: P O Box 27-103, Wellington, New Zealand