Subj : not ignoring you... but...
To   : mark lewis
From : Scott Little
Date : Sat Jul 10 2004 08:13 am

 [ 09 Jul 04 13:44, mark lewis wrote to Scott Little ]

 SL>> Gawd DAY-EM!  What a rats' nest.
 ml> where? in ele or the mk library?

Ele.  The ifdefs.... awmigawd the ifdefs...  *cries*

 SL>> odd since a simple unix2norm(norm2unix()) would tell you it's
 SL>> broken,

 ml> hummm... wonder what that compares with in the mkmsg stuff? mkmsg uses
 ml> UnixToDT and DTToUnix to convert the datetime record format back and
 ml> forth...

The mkmsgjam.pas in the Ele source tree has been modified to use unixdate.pas 
and jdates.pas from the parent (elebbs) directory. In fact, the whole mkmsg 
source included with Ele has been drastically pruned.

The original mkmsg source (v1.04 is what I have) uses a different algorithm.  
I'll have a look at it later and see if it's accurate, if so I'll massage it 
into unix2norm() form, since it's no doubt much more tested than my code.

 SL>> and this is a piece of code that's been floating around for
 SL>> ages (and obviously used by many people).
 ml> makes me even more curious, now...

It's from SWAG, BTW.  And the first whole page of hits off Google for any 
searches mentioning unix, dates, and Pascal.

 SL>> The fix eludes me (at 5am), so I rewrote it.  This isn't
 SL>> well tested, but it Works For Me.
 ml> did you test leapyears ;)

Of course :)

It doesn't take into consideration leapseconds or any of that other stuff 
though.  But a few seconds here and there over a couple of decades is accurate 
enough for our purposes.


-- Scott Little  [fidonet#3:712/848 / sysgod@sysgod.org]

--- GoldED+/W32 1.1.5-31012
 * Origin: Cyberia: 100% Grade "A" mansteak baby. (3:712/848)

.