Re: [ft] Large delta_y for hinted vs unhinted "COMBINING CARON BELOW" glyph

[Nigel Tao]

On Thu, Oct 31, 2013 at 12:06 AM, Werner LEMBERG <address@hidden> wrote:
>> Another clue: the ttx file lists an hmtx table but no vmtx table...
>
> This is completely irrelevant to the problem.

Ah, I was mislead because the SRP0 instruction sets RP0 to point to
point number 10, which is the 4th phantom point (the caron has 7
points), and the 4th phantom point's Y co-ordinate is:

(loader)->top_bearing + (loader)->bbox.yMax - (loader)->vadvance.

I guessed that the top_bearing and vadvance were calculated
incorrectly if the vmtx table wasn't present, but on second thoughts,
the more plausible hypothesis is that the phantom point used for
hinting the outer glyph (#733) is (incorrectly??) not reset from the
phantom point used for the inner component (#649).

The bbox.yMax for #733 is -116, for #649 it's 1638, and that
difference in bbox.yMax values would explain the eventual vertical
difference in hinted glyphs via the ALIGNRP.