[GRASS5] datum shifts

[rgrmill at rt66.com]

Folks,

I have upgraded the proj library in my installation to version 4.5.5--it 
may actually have been at that level to start with, I don't know.  I 
added some code to the library and modified r.proj, v.proj and s.proj to 
perform the nad27<->nad83 datum change in a pretty seamless manner.  For 
our North American users, I'd like to get this capability into 
GRASS5.0.0 before the feature freeze.

For now the datum shift is limited to the changes between the two nad 
data.  I hope the mechanism is sufficiently general that other datum 
shifts can be added with relatively little change.

I need some help to get the changes installed in GRASS before the 
feature freeze.  My biggest problems are:

The nad27<->nad83 conversions are based on data tables.  The data tables 
are shipped in ascii form.  They have to be interpreted to binary by a 
program that ships with the package and then placed somewhere in the 
$GRASSBASE hierarchy.  Currently I have them in $GRASSBASE/etc/nad.  It 
would make more sense to me to put them in $GRASSBASE/lib/nad.  Your 
input would be appreciated.

The user modules must know where the data tables are stored, and that 
information should be compiled into the modules.  I think that would be 
done by configure, but I don't know enough to say, much less enough to 
make the change myself.

I don't know how to go about doing either of those tasks.  Help?

I will also have to commit and/or add the library code.  I think I can 
figure that out.

There are some outstanding tasks.

I probably need to get the proj library list of projections in sync with 
the GRASS 5 list.

At this point the modules are hardwired to perform the nad conversion 
only in the conterminous US and adjacent areas and only at regular 
precision.  That needs to be changed.  I would need either to install 
the ability for the modules to determine which table they should use for 
the input area or I need to make that an argument.  Using the high 
precision tables (maybe not necessary) would require another argument.  
Added arguments make the datum change less seamless, so users outside NA 
might find that annoying.

I have not upgraded m.proj to complete the datum change.  I don't regard 
that as a high priority.

Your input and help would be much appreciated.


Roger Miller