1993.06.21 21:07 "Re: TIFF - T4 2D subtleties", by Sam Leffler
In implementing Group3, 2D, TIFF files, there seems to be confusion concerning the positioning of the EOL+1 and EOL+0's relative to byte boundaries, strips (and tiles).
TIFF6 says that fill bits, if added, are such that EOL always ends on a byte boundary.
I think tifflib interprets the spec by ensuring that EOL+1 or EOL+0 always ends on a byte boundary. Doesn't this contradict the spec?
Secondly, does a strip (or tile) always commence with EOL+1? If so, the last row in a strip does not contain an EOL, only fill bits, so up to 10 fill bits may occur before the EOL. Or should the 1 be at the start of the strip (except for the first strip, which starts EOL+1)?
Any views? Or is it formally defined somewhere?
At one time the library aligned the bits such that the tag bit that follows the EOL code was included in the byte-aligned data. This was wrong and was fixed in February of last year.