Date:Fri, 7 Jul 2006 13:08:44 -0400
Reply-To:Subject Coordinates Discussion List <[log in to unmask]>
Sender:Subject Coordinates Discussion List <[log in to unmask]>
From:Archie Warnock <[log in to unmask]>
Organization:A/WWW Enterprises
Subject:Re: The proposal passed MARBI!
Comments:To: Subject Coordinates Discussion List <[log in to unmask]>
In-Reply-To:<[log in to unmask]>
Content-Type:text/plain; charset=ISO-8859-1
I believe the decision regarding sexagesimal vs. decimal depends on the
primary purpose for which the coordinates are being inserted into the
records. If they are to be scanned by (human) eye for comparison
purposes, D/M/S is probably preferable. In my experience, it's easier
to eyeball a sense of relative scale, or do manual error-checking,
between values if they're presented that way.
On the other hand, given that MARC is largely used as machine-readable
records, doing conversions in and out of sexagesimal require extra
computations (admittedly almost trivial) but also raise the possibility
of conversion errors. You pays your money, you takes your chance.
Precision in the computer representation of floating-point numbers is a
long-standing bugaboo (cf, the FITS format used for exchange of
astronomical data), but decimal degree notation shares that problem with
the seconds part of D/M/S coordinates unless one is satisfied with 1"
accuracy. Remember, only integers are inherently exact once they're
loaded into the machine.
Ultimately, it seems to me that using decimal degrees is more in keeping
with the most likely uses of spatial coordinates in the authority records.
If the need for precision is really critical, it would be possible to
use scaled integers to represent decimal degrees, albeit requiring an
additional value to hold the scale multiplier for each coordinate.
Just my thoughts from out here in the spatial metadata world...
--
Archie
-- Archie Warnock [log in to unmask]
-- A/WWW Enterprises www.awcubed.com
-- As a matter of fact, I _do_ speak for my employer.
