New to Typophile? Accounts are free, and easy to set up.
Bogus FontLab error = table overflow?
Posted by Thomas Phinney in
14 Apr 2011 — 8:05pm
14 Apr 2011 — 8:05pm
I get the interesting error below when I try to compile features in an OT script font. The notable things are that the claimed number of elements is absurdly large, and the error message is claiming that a number is not equal to itself. The only features in the font are calt and maybe kern.
It seems to me that I have seen this error before, and it does mean something, but as the error message itslf is bogus, I can't recall for sure what that is. Is it related to overflow of the 64K size limit on a subtable in OpenType? That rings a bell for me....
[ERROR] Target glyph class in sub-rule doesn't have the same number of elements as the replacement class; the target has 87747800, the replacement, 87747800 [//vmware-host/Shared Folders/My Documents/Shared Data/FontLab/Studio 5/fontlab.fea 133]
User login
Support our Sponsors
14 Apr 2011 — 11:32pm
Of course, another possibility is that although the information about the number of elements is bogus, there really is a mismatch. This is probably the most likely possibility....
T
15 Apr 2011 — 12:52am
If you save out the features file and open in a text editor, is there anything on line 133 that might provide a clue?
15 Apr 2011 — 7:52am
Whenever I get errors of this type in FL5, the numbers reported are *always* wrong, but the error itself is real. I'd definitely check that possibility first.
André
15 Apr 2011 — 8:33am
Thanks André. :)
It is distinctly possible that there is a mismatch. We'll dig into it.
Cheers,
T
15 Apr 2011 — 8:45am
I remember getting this error a few times. I don't recall the REAL reason but I do recall that the error message had nothing logical to do with the solution. It may have something to do with kerning classes but I can't be sure. Try rebuilding the kern feature from the "Kerning Assistance" window and then save quit, and restart FontLab.
16 Apr 2011 — 11:25pm
Actually did turn out to be what it claimed... just had to ignore the bogus parts of the error message.
Cheers,
T