These are chat archives for caryll/otfcc

10th
Apr 2017
Cosimo Lupo
@anthrotype
Apr 10 2017 12:16
@be5invis should textOffset in TSI0 be interpreted as a signed or unsigned 32-bit integer? I see otfcc reads it as unsigned, but fonttools as signed. What does the spec say about it?
Belleve Invis
@be5invis
Apr 10 2017 12:18
uint32.
Cosimo Lupo
@anthrotype
Apr 10 2017 12:18
excellent, thanks!
Cosimo Lupo
@anthrotype
Apr 10 2017 12:44
can the sequence of (glyphID, textLength, textOffset) in TSI0 be discontinuous and contain gaps, e.g. if an entry contains no data? Or should there be an index record for every glyph in the font (len(indexPart) == maxp.numGlyphs) even when a glyph has no VTT program defined?
fonttools assumes that TSI0 always contains maxp.numGlyphs + 5 number of indices
the 5 includes the "magic", plus the four extras
Belleve Invis
@be5invis
Apr 10 2017 12:45
Items must be continuous. though some data can be empty (i.e. textLength = 0)
Cosimo Lupo
@anthrotype
Apr 10 2017 12:46
ok, thanks
it sounds like a waste
Cosimo Lupo
@anthrotype
Apr 10 2017 13:47
are the four "extra" programs always there, even if empty?
I mean the index referencing the four extra programs in the TSI0 table.
Belleve Invis
@be5invis
Apr 10 2017 13:48
yes. always there.
leave textLength = 0 for empty.
Cosimo Lupo
@anthrotype
Apr 10 2017 13:48
great, thanks for the quick reply!
ok