[XeTeX] Strangeness with Minion Pro Opticals Small Caps

[Jonathan Kew]

On 21 Aug 2006, at 11:52 am, Peter Heslin wrote:

> Jonathan Kew <jonathan_kew at sil.org> writes:
>
>> But the behavior he's seen is completely consistent with having
>> incorrectly ordered lookups in the GSUB table, which is a plausible
>> error
>
> Oops -- mea culpa.  I've tracked down the problem -- it turns out  
> that I
> had modified the font a while ago.

Aha! So that's it.... it hadn't occurred to me that this was a  
possibility, but of course it explains the discrepancy.

> This script only adjusted the left bearing of a few Greek glyphs,  
> so it
> should not have touched the GSUB table.  It must have been my own  
> error
> while developing the script or a bug in FontForge that has since been
> fixed, since re-running the script now on the original AR-supplied  
> font
> does not result in a font with a similarly corrupted GSUB table.

I'd expect FF to rebuild the GSUB table (and others) when you  
generate the font, even if you aren't aware of having touched  
anything directly related to it. Remember that it refers to glyphs by  
index, so if there's even a possibility that the glyph palette has  
changed (been reordered, glyphs added/removed, etc) then the old GSUB  
will be invalid. So a tool like FF most likely interprets it on  
import (resolving glyph IDs to names, etc) and regenerates it on export.

So it's very possible that a FF bug could lead to the GSUB table  
being incorrect. And indeed, the FF changelog at sourceforge includes  
an entry (sometime between April and July) that mentions "lookup  
ordering within GSUB has been totally broken.....". That's probably  
why it went wrong; and you've since acquired a fixed version.

JK