[GRASS5] [bug #1065] (grass) GRASS and IDRISI

[Eric G. Miller]

On Mon, May 27, 2002 at 01:08:44PM +0200, Request Tracker wrote:
> this bug's URL: http://intevation.de/rt/webrt?serial_num=1065
> -------------------------------------------------------------------------
> 
> > On Wed, Mar 27, 2002 at 05:49:34PM -0700, David A Hastings wrote:
> > > I hope to set up the database so that source (e.g. PERMANENT) data will
> > > be readable on both GRASS and Idrisi.  GRASS and Idrisi 8-bit
> > > uncompressed cell files are identical, one can use links to headers, so
> > > that source data are interoperable on both systems.  I have been doing
> > > this since 1995.  If GRASS let one specify an actual 2-byte integer, and
> > > specify the byte ordering (say as enhancements to CELLHD files), as is
> > > possible in programs such as Photoshop, the interoperability would be
> > > better.  And Idrisi has had floating point cell files since its
> > > inception - though I won't know about compatibility issues with GRASS 5
> > > for another few weeks.  Finally, if GRASS could let the user force files

I suspect GRASS and Idrisi FP formats are significantly different.  For
FP rasters, GRASS actually stores the data in a different file and has
an additional header file.  Whether float, or double size, each file has
an index of row offsets (int) followed by the data.  The data itself is XDR
encoded, and each row may be compressed via zlib.  The first byte of
each row flags whether the row data is compressed ('0' - no, '1' - yes).

GRASS5 also has a NULL data file to differentiate 0 from NULL.  This is
used with all raster types.

-- 
Eric G. Miller <egm2 at jps.net>