| Re: Map Projection Doom and Gloom [message #66672 is a reply to message #66634] |
Fri, 05 June 2009 08:40  |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
David Fanning writes:
> Here is the crux of my advice: If you are going to use
> a map structure for *ANYTHING* (and I really do mean
> *ANYTHING*!!), then be absolutely sure you execute
> the MAP_PROJ_INIT command *IMMEDIATELY* prior to using
> the friggin' thing!
Just to follow up on this. I've heard from someone who
understands exactly what is going on here. It seems that
the underlying GTCP software that IDL relies on for it's
map projections uses static variables (explained to me as
"equivalent to evil common blocks") to hold various parameters.
(If you look in the MAP_PROJ_INIT source code, you will
see that the ITTVIS programmer puts a helpful note in the code
that the user is calling the GCTP software "at your peril".
Not the kind of thing you really like to see in software
comments.)
So, while the *structure* returned from MAP_PROJ_INIT
doesn't actually change when another MAP_PROJ_INIT call is
made, the underlying data stored in these static variables
*does* change. This is what causes the Great Disconnect
as you compare an unchanged map structure with wacky output.
It is unclear if this is an IDL problem or a GCTP problem.
ENVI, I note, has given up on GCTP completely and is going
(really *unfortunately*, IMHO) to an ESRI map projection engine.
It would have been SO much better if they just moved all this
map projection stuff to a proj.4 Open Software project. (The
ESRI engine uses *some* of the proj.4 stuff, but not all of it
for some reason.)
There is no indication at all that IDL's map projection
software will change any time soon, no matter what ENVI
does. So I wouldn't hold my breath that this will be fixed
in our life times. :-)
Cheers,
David
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
