[Gdal-dev] RE: MapServer WCS

[Martin Daly]

Frank,

> I have used PLANARCONFIG_SEPERATE within GDAL (and PCI's GDB 
> before that) by default because it is more efficient for 
> access given the band-by-band approach of GDAL.  This makes 
> some sense when GDAL is used to produce TIFFs for consumption 
> in GDAL, but it is true when aiming for maximum 
> interoperability it would be better to use PLANARCONFIG_CONTIG.
> 
> Steve ... I would suggest you add the FORMATOPTION 
> "INTERLEAVE=BAND" line in the OUTPUTFORMAT declaration in 
> your mapfile to improve interoperability of the TIFF files.
> 
> Martin ... I would encourage you to support 
> planarconfig_seperate files in your built-in TIFF reader if 
> possible.  They aren't too hard ... just yet another case.

What I would like to do is support these files (and some other more
esoteric TIFFs) using the GDAL Plug-in for SIS.  But I'd like to keep
our current internal support as well, for backward compatibility and
other reasons.  SIS offers files to Plug-ins before attempting to use
them itself, so what I'd like to be able to do is allow the GDAL Plug-in
to determine whether or not to accept the file.

All of which is a long-winded way of saying, what do you think about
giving public access to the GTiffDataset hTIFF member (it is currently
private)?  This way I could interrogate the underlying TIFF file to see
if it was, for example, PLANARCONFIG_SEPERATE.  My nuclear alternative
is to use libtiff directly, but I'd end up having to open the file using
GDAL as well just to check that GDAL was happy with it.

Regards,
Martin