2012-12-14 15:48, Andreas Prilop wrote:
>>
http://www.cs.tut.fi/~jkorpela/test/dotop.html
[...]
> Include U+00B7 and U+2219 in your test.
Done.
> It may well be that IE
> uses the same glyph for all three characters.
It doesn’t, but you need to zoom in to see the differences in Cambria.
> What do you get in word processors?
In Word 2007, U+22C5 in Cambria turns to a question mark in a rectangle
when I bold the text. In LibreOffice, it becomes bolder.
Now I realized I should take a look at the Firefox rendering using the
Font Information add-on. It shows the problem glyph as Cambria, not as
Cambria Bold as it does for real bold Cambria. This is odd because the
glyph is clearly of bold design.
The analysis is confirmed by looking at the Cambria Bold font in DTL
OTMaster Light: Cambria Bold lacks any glyph for U+22C5.
So apparently browsers just apply algorithmic bolding to U+22C5 in
Cambria, and the problem can be identified as a Firefox bug in
implementing such “fake bolding”. It seems that for some odd reason,
Firefox changes the character to U+1E61 and then bolds the glyph (that’s
at least what I gather from Font Information, but to my eye, the result
looks just the same as U+1E61 in Cambria Bold – but maybe algorithmic
bolding works that way in this case?).
--
Yucca,
http://www.cs.tut.fi/~jkorpela/