Perhaps the vertical metrics. Is the font's UPM 1000? Have you ran the font through CompareFamily?

That looks like the caret height and refresh problem resulting from a CFF font having a UPM other than 1000. But I thought that had been fixed in CS3?

The UPM is 2048. I haven't run compare family. I will do that tomorrow.

Okay, just had a look. Compare family gives a 76 KB serving of fail. I had a look with fontQA, and this is what is says:

Vertical Metrics
Test Message Details
Y-min Family Statistic min: -497, max: -497, rng: 0, avg: -497, sdv: 0
Markant Y-min -497
Y-max Family Statistic min: 2080, max: 2080, rng: 0, avg: 2080, sdv: 0
Markant Y-max 2080
UPM Not all UPM values are matching the required value 1000
Markant UPM 2048
equal Ascender All Ascender values are equal. 1490
Markant Ascender 1490
equal Descender All Descender values are equal. -393
Markant Descender -393
OS/2-sTypoAscender All OS/2-sTypoAscender values are matching the calculated value
Markant The OS/2-sTypoAscender value matches the required value OS/2-sTypoAscender: 1490
OS/2-sTypoDescender All OS/2-sTypoDescender values are matching the calculated value
Markant The OS/2-sTypoDescender value matches the required value OS/2-sTypoDescender: -393
equal OS/2-sTypoLineGap All OS/2-sTypoLineGap values are equal. 574
Markant OS/2-sTypoLineGap 574
OS/2-usWinAscent All OS/2-usWinAscent values are matching the calculated value 2080
Markant OS/2-usWinAscent 2080
OS/2-usWinDescent All OS/2-usWinDescent values are matching the calculated value 497
Markant OS/2-usWinDescent 497
hhea-Ascender Not all hhea-Ascender values are matching the recommended value
Markant The hhea-Ascender value does not match the recommended value hhea-Ascender: 1490, recommended: 2080
hhea-Descender Not all hhea-Descender values are matching the recommended value
Markant The hhea-Descender value does not match the recommended value hhea-Descender: -393, recommended: -497
suspect BBox height The BBox height of all glyphs is in a normal range.
Markant The BBox height of all glyphs is in a normal range.

I've just tried Adobe Arabic (which is an OT-CFF font with a 2048 UPM) in InDesign CS4 and saw no selection issues.

Claus, run this command:
spot -t head YourFont.otf | grep unitsPerEm
What result do you get? unitsPerEm=2048?


BigMac:Markant claus$ spot -t head Markant-Regular.otf | grep unitsPerEm
unitsPerEm =2048


And damn Typophile won't let me insert an image now, otherwise I could show the vertical metrics from fontQA. That is why I had to paste that illegible mess in the post above.

I think I found the problem. I'm assuming that you're generating the PFA file from FontLab Studio (v5.0.4), yes?

When the font's UPM value is different from 1000, FontLab Studio generates the PFA file with incorrect FontBBox values. The FontBBox values are correctly displayed in the "Key dimensions" pane of the Font Info dialog; but the values in the generated font will be different.

For quick workaround, just open the PFA file in a text editor and edit the FontBBox values to match what's displayed by the Font Info.

BTW, CompareFamily quickly pointed me to the reason of the problem. Single Face Test 15 (Warn if Font BBox x/y coordinates are improbable, or differ between head table and CFF) output the following message:
The font bounding box in the hhea table '(-322, -512, 2306, 1798)' differs from that in the CFF table '[-157, -250, 1126, 878]'.

Hi Miguel
Thanks a lot for looking into this. It halfway worked, now the rendering updates as it should. The only thing is that the selection still looks like before – that is only colouring from the x-height and down to somewhere below the baseline. This is a minor issue, so let's not waste time on pursuing that.

Claus