[css2] New data for issue 95

--HG--
extra : convert_revision : svn%3A73dc7c4b-06e6-40f3-b4f7-9ed1dbc14bfc/trunk%402385

Author: Hixie

css2/issues/issues-3

@@ -2188,6 +2188,24 @@ following paragraph now in the spec:
    glyph measurement if possible, but may fall back on a default of 0.5em
    if the first available font does not contain suitable information.
 
+Ian responded to the second of Bert's ideas above with:
+
+> This would work too. It differs from the alternative proposal in that
+> it uses the 'ex' height of the first font _with an x_, rather than
+> the first available font. Is that what we want?
+
+To which Bert replied:
+
+That's what I gathered from the minutes:
+http://www.w3.org/2005/10/17-css-irc.html#T22-09-17
+
+I think the argument was that, if you use an ex and there is an x in the
+content, you would want the two to match. Plus, UAs already know how to
+search the font list for the font with an x.
+
+I think that's reasonable, and it made the algorithm nice and short. But
+I'm open for suggestions.
+
 =WG= Discuss.
 ------------------------------------------------------------------------
 Issue 96.