NetCDF / OPeNDAP viewer prototype: VuNcDAP.exe

VuNcDAP.exe will do:

	Display header info (equivalent to CDL)
	Display geographic and time coverage, if readable.
	View a scalar item and single pair of vector components
	Subset a scalar item and single pair of vector components into a local nc file

Installation:

	Install first OPenDAP API package (see note at).
Copy VuNcDAP.exe and VuNcDAP.ini to your PC. Run VuNcDAP.exe.
Depending on location of the executable, paths must be recorded in VuNcDAP.ini file. On initial attempt, program asks paths of,

	1. Base coastline 
	2. Notepad.exe (or equivalent)
	3. Initial sample NC file

Default paths for these are (or see/edit manually VuNcDAP.ini);
 
	1. Base coastline 			>app.Path & "\coasts\world_W.bdm"
	2. Notepad.exe (or equivalent)	> winnt\system32
	3. Initial sample NC file		> app.Path & "\local_nc\*.nc"

Menu:

Load	Loads a single file. "?" triggers the Command dialog to look for a file in local path
	For URL, you must get it somewhere else and paste into the load area

Load2	Loads a pair of files that contains only one of the vector components each. Left 			command ">" will copy text of "load area 1" to "load area 2", then edit file name 			to make a pair.

	When the loading is successful, coordinate variable (lon/lat) will be displayed in the 	preview pane with Green/Blue/Red color-wise as indexes increases.

List CDL Header
	List dimension, variables, attributes in CDL format (using NotePad.exe)

Plot to PicBox
	Individual item may be viewed in a spatially windowed /sub-sampled manner

Subset to local
	Items selected can be saved in a local nc file


 
Program Note:

Actually there are no differences in working with local nc files versus URL files (http//); codes you already have will work just fine. 
 
Strategy to get contents of URL files
Depending on where a URL file is located, accessing data contents is very slow. It seems each transaction has a significant overhead. For given an amount of data, less transactions result less overall time, if you can endure black out time in which you have no way to stop.
Typical size of a block data may be A(time,level,lon,lat), where time=1, level=1, lon=35, and lat=100, whereas in making lon=1/lat=100, and repeating requests for 35 times gives you more control but will take much long time to finish. This would be mute point if a retrieving is done in background task. Although, knowing how much and how long it would take may be critical how to handle this process.

Presently, VuNcDAP accesses (limits) URL contents by spatial windowing and sub-sampling, and also with only one time and one level for the plotting operation. When you wan to locally-save the same contents, probably you will not do sub-sampling and take all time steps available. It will be always the case that you can not afford to take all contents, as some files are as large as ~several hundreds Meg.  

Major works involve to distinguish unknown variable type (long, float double, etc) and the dimension order (,,lon,lat) versus (,,lat,lon), also to handle a pair of files that contains vector components in separate file.
