Bob, about a version of Guide on DVD: had I realized Guide 8 was
going to take so long, I might well have gone that route. It not
only would evade disk-swapping; it would let me store _lots_ of data.
DVDs are getting sufficiently cheap that it wouldn't surprise me if
some people would purchase them just to be able to run a data-heavy
version of Guide on them.

Laurent, as to the two bugs you found: I've revised Guide so that
Saturn's rings don't vanish simply because the globe of Saturn itself
is occulted, and have fixed the "polygon" appearance of the edge.

You should expect accuracy to be good down to about the one-second
level. At that point, there are three significant sources of error.

One problem arises because there are several different variants of
UT. When Guide says "UT", it essentially means UT1, the "freely
flowing" flavor of Universal Time. In UT1, there are no leap seconds;
if the irregular rotation of the earth leads to a longer "day" than
usual, each UT1 second is expanded a little so that there are still
24 * 60 * 60 = 86400 UT1 seconds during that rotation.

UTC (Universal Coordinated Time) is the variant in more common
use. In UTC, seconds are fixed in length; if the earth's rotation
happens to be a little more than 86400 seconds, the error accumulates
until eventually, an extra "leap second" is inserted (at the end of
either June or December). As a result, UTC and UT1 can differ by up
to a full second.

The second problem crops up because Guide uses ELP for lunar positions,
and that's not accurate to much better than a second. (Incidentally,
it's well to use 'full precision' when you're trying for this level of
accuracy.)

Finally, you've got the fact that Guide shows the moon as a sphere.
The moon moves along at about a kilometer a second. If there's a
one-km elevation difference from the mean lunar radius at the site in
question, it'll throw off Guide's result by that amount (more if the
contact is at a shallow angle).

I may rectify the first problem by including tables/formulae to compute
UTC-UT1, the second by using JPL DE ephemerides instead of ELP. Both
fixes are real possibilities once Guide 8 is shipped. The third problem
would actually not be all that difficult to solve if I had accurate
elevation data for the moon. I can get such data for the Earth, Mars,
and Venus. It apparently does not exist for the moon. Until it does,
this one isn't gonna happen.

-- Bill