[GRASSLIST:116] Re: 5.0.0pre3 to 5.0.2 conversion problem

[Roger Bivand]

On Wed, 21 May 2003, Michael Ash wrote:

> Dear Mike
> 
> Thank you for the interest in my difficulties.
> 
> 
> > What happens when you substitute "x0" for the "PNG" driver?
> 
> The PNG driver is not the source of the problem.  I tried the script
> with d.mon start=x0, etc., instead of PNG and had the same problems.
> 
> > What computer/CPU/OS?
> 
> Intel (Pentium II or something) Dell Inspiron 4100/Windows XP/Cygwin/xfree86
> 
> > No data format changes should have occurred between those two versions.  I'm
> > wondering whether there is a difference in the region or projection/datum
> > handling leading to attempts to retrieve data out of the region.
> 
> GRASS 5.0.2 displayed maps that looked about right in terms of the
> geographic layout.  Resetting the region worked as I expected, etc.  
> The raster values seemed out of whack: in one raster map, the values
> were right according to d.what.rast but the non-null cells were all
> white instead of color-graded; and the values were all wrong (huge
> negative numbers, long strings of numbers) in another raster map; in
> most of the maps the color scheme had changed.  (It almost looked as if
> GRASS was changing datatypes.)

Just a thought - I can date the lzw to z compression change to 25-Nov-2000 
(the r.lzw2z README) - what date was 5.0.0pre3? I recently hit an old 
raster still in lzw compression, and puzzled a bit before remembering the 
shift. I think 5.0.0pre3 was early 2002, so having an lzw floating-point 
raster isn't likely (and probably wouldn't open), but it may be worth 
checking when it was created.

> 
> > Have you tried recreating the project with the new version of Grass.
> 
> This is an excellent suggestion, and I have not, but I will.  The
> project is computationally intensive (lots of smoothing); so I need to
> set aside time for me and the computer.
> 
> I have done new work in 5.0.2 since you posted it and everything has
> been a-ok.
> 
> 
> Thanks again for your interest and suggestions.
> 
> Best,
> 
> Michael Ash
> mash at econs.umass.edu
> 
> 

-- 
Roger Bivand
Economic Geography Section, Department of Economics, Norwegian School of
Economics and Business Administration, Breiviksveien 40, N-5045 Bergen,
Norway. voice: +47 55 95 93 55; fax +47 55 95 93 93
e-mail: Roger.Bivand at nhh.no