Thanks for looking at it Yoshiki, I'll follow your suggestion. It looks 
like that is what StrikeFont does. (With baselineY being ignored, hmm, I 
guess thats ok for single byte strings?...)

In the long run, it seems like the primitive could be extended to handle 
multibyte strings too. The mechanics seem pretty similar and you don't 
want a noticeable speed difference between the two if it can be avoided.

Eddie

Yoshiki Ohshima wrote:

>  If you really like to get that kind of performance improvement, you
>could insert a class test in:
>
>StrikeFontSet>>displayString:on:from:to:at:kern:
>
>so that it would look like:
>
>	(aString isMemberOf: ByteString) ifTrue: [
>		...
>	] ifFalse: [
>		^ self displayString: aString on: aBitBlt from: startIndex to: stopIndex at: aPoint kern: kernDelta baselineY: aPoint y + self ascent.
>	].
>
>or something.
>
>  I haven't tested yet, but probably better than just replacing it.
>
>-- Yoshiki
>
>
>  
>