1999.06.10 13:52 "Re: Large File Support", by Frank Warmerdam
We need a real honest-to-goodness *standard* (without options) so we can write good (and freely available) software. While standardizing the format it would also be an excellent idea to standardize the programming interface to read/write/verify/info-extract libraries.
I disagree. I think it's extremely important that I be able to implement my code and API completely differently from yours and be able to call both of our libraries standards-conforming so long as we both pass a test suite. I don't want to be prevented from claiming to be a compliant TI64 producer or consumer simply because I used C++ when the standard API was specified in C...
I agree that the data format definition must stand on it's own without the requirement to use a particular library implementation or even API. On the other hand, I think it can be a big help folks who want to follow the standard properly to provide a reference library implementation of the standard and that by all means we should do this.
However, I am still concerned that a TIFF64 standard, or something similar, would never gain wide enough adoption of be of much use. I might achieve some success at getting a new TIFF format adopted in the GIS/RS community but to get desktop photo editing packages to adopt it would be alot harder.
I think a large part of the utility of TIFF format in the GIS/RS field is that it provides data sharing with non GIS/RS packages such as photoshop. If we don't achieve that breadth of adoption in a new standard then it is pretty well pointless.
We might be better off defining an extention whereby a TIFF file can refer to a series of other tiff files which together form a larger image. Then each file could be less than 4GB, but the total image could be more. Packages which don't understand the ``assembly of TIFF files'' tags could still operate on each sub-TIFF file independently. libtiff could be modified to automatically assemble the set of files into one virtual image through the standard APIs.
Would such an approach be of any interest?
I set the clouds in motion - turned up | Frank Warmerdam, firstname.lastname@example.org
light and sound - activate the windows | http://members.home.com/warmerda
and watch the world go round - Rush | Geospatial Programmer for Rent