Success (sort-of)!

[Russell Nelson]

Success!!  I got my dataset to display correctly!

That's good for me in that I can get my task done.  However, the
trouble that I had indicates that Thuban is fragile.  It doesn't
withstand native users very well.  It has, in essence, the same
problem that GRASS has: intolerance of ignorance.

Here's what I had to do:
  1) start with a new session (by running Thuban)
  2) set the projection of the session using Map/Projection...
  3) load the raster map, and set its projection using Layer/Projection...
  4) load the shapefile.

How can we change the code so as to prevent other people from having
the same trouble I had?  (Not to mention preventing the resultant zero
scale problem.)

We could do it heuristically: by noticing that, when someone loads two
layers, the unprojected layers don't overlap, but there is a
projection which causes them to overlap.  Emit a warning that says
"Select <LAYER> and use Layer/Projection..."

We could do it algorithmically: by refusing to display anything if two
layers are loaded and any of them remain unprojected.

.... or by setting the Map projection (if None) to default to the
projection of the first layer loaded.

Yes, I know that all of this is documented.  The software shouldn't
misbehave if the user doesn't read or obey the instructions.

-- 
--My blog is at angry-economist.russnelson.com  | Freedom means allowing
Crynwr sells support for free software  | PGPok | people to do things the
521 Pleasant Valley Rd. | +1 315-323-1241 cell  | majority thinks are
Potsdam, NY 13676-3213  | +1 212-202-2318 VOIP  | stupid, e.g. take drugs.