Re: rasterization weirdness with Skurri.TTF?

[Just van Rossum]

At 2:00 AM +0100 03-11-2000, Werner LEMBERG wrote:
>> I'm looking at Skurri.TTF
>> (http://rover.wiesbaden.netsurf.de/~kikita/s.htm) and am having
>> trouble understanding some of the metrics that are returned.
>
>This font is buggy; the `loca' table appears corrupt to FontTools, and
>`hmtx' contains too much data -- Just, can you check this?  Maybe you
>can implement some recovering so that ttDump doesn't abort but emits
>warnings instead.
>
>> First of all, the yMax and yMin values for the face's bounding box
>> are 988 and -306, respectively. That would mean that the tallest
>> glyph for the entire face would be 1294/64 (or 20.22) font units
>> tall.
>
>Don't divide by 64 -- font units are integer values.
>
>> However, as I'm loading glyphs from the face, I see that the 'l'
>> (lowercase L) glyph (among others) has a height of 1794/64 (28.03),
>> which is far greater than the bounding box height of the entire
>> face!
>>
>> I can see why someone would design a font that's greater than the
>> height (asc+desc+lead) of the face, but why would the face's
>> bounding box not report this height?  Isn't the bounding box
>> supposed to enclose all glyphs? It seems weird to me that it isn't
>> consistent with some of the letters in the face.
>
>A lot of data seems to be incorrect...

It's the loca table and the hmtx table. I've just checked in TTX fixes that
make this font go through all the way.

Just