1999.06.09 18:50 "Re: Large File Support", by Chris Barker
Glen, I think your idea has merit, and something like this would probably be the right way to go for any TIFF library.
However, we first need a specification that describes the details of the new format. This new format would _have_ to have the "magic" number changed since the initial IFD offset will be 8 bytes instead of 4. If we don't change the magic we will break all existing TIFF readers.
Could we call this TIFF? I believe that Adobe has the name trademarked or copyrighted (whichever is appropriate). Could we use all the ideas in TIFF (tags, tag names, etc..) without Adobe's blessing? At the start of the TIFF 6.0 document there is some info on how to request changes to TIFF, but I have had no success trying to get Adobe to pay attention to me.
If we change the format enough that the existing libraries cannot be easily adapted to supporting the old, and new formats I think we should ask ourselves whether the world needs yet another image format.
It seems to me you have not been paying attention. The world *definitely* needs another file format that can support files larger than 4Gb, as evidenced by all this email! What we don't need is another proprietary format.
Assuming that we could get away with calling it TIFF64, and that libtiff could be adapted to supporting the old and new format I think it would be world pursing a 64bit version of the format.
I would not suggest keeping libtiff as it is. We need public domain code along with a public domain format, but what we *really* need is a standard interface spec for the parse/write library to go with the file format. This way we can have public and proprietary implementations that are interchangeable...
I would hope we could do this in a way that other TIFF reading libraries could be fairly easily adapted to also support the TIFF64 standard. I think this should be one of our objectives. Of course, keeping essentially the same model with modest changes in the format to support larger numbers should go a long way to satisfying that objective.
Since we have to change, I believe we should change to improve where possible; but I think TIFF64 (or whatever) should be simpler if possible. The TIFF model has much which is good, but there are way too many possible variations that are not specified. That should be fixed.
Barker's law: For every inaction, there is an equal and opposite excuse.
Check out Interlinear's web page at http://www.ilt.com