Stan Brown <th************@fastmail.fm> wrote:
"Barry" <bg***@yahoo.com> wrote in
comp.infosystems.www.authoring.html:I've noticed a strange error on my website. When I print a capital
letter P with a dot above, using & #7766; it appears correctly, but
when I use P& #0775 it doesn't.
I've presented the problem here http://baz.perlmonk.org/example.html
t's exactly the opposite for me: your four "775" examples all have
the dot above, and your other four examples have the dot looking
like a period (full stop) after the letter.
On my IE 6 (Win), I see the same situation as Barry: Ṗ gets
displayed incorrectly, other cases are OK. (Barry probably mistyped the
character references here to prevent newsreaders from interpreting.
But if someone's newsreader interprets strings as character references,
or as HTML markup in general, <blink><font color="red" size="7">the
newsreader needs to be fixed</font></blink>. Misrepresenting data is not
a solution. And a leading zero, though allowed, is rather pointless in
decimal character references.)
Specifically, I see Ṗ as capital P followed by a small box. The box
is IE's way of telling about a character it cannot render. I have no idea
why this happens, but we know IE's mechanism for dealing with combining
diacritic marks is defective. It can handle simple cases but fails often
miserably. However the usual failure is putting the diacritic at a wrong
vertical position, because IE does not pay attention to the height of the
base character - it probably just overprints the base character with a
glyph of the diacritic in a fixed position.
On the practical side, precomposed characters like Ṗ often work
better than decompositions like Ṗ, because
a) they work even on browsers that have no support to combining diacritic
marks but are able to use a rich enough font (e.g., several older
versions of IE)
b) when they work, they generally produce a better visual presentation,
since the glyph is designed by a typographer, instead of being
produced by a (simplistic) program that combines two glyphs.
On the other hand, it is possible that the font in use lacks a glyph for
the precomposed character but the browser would be able to handle the
decomposition using its general algorithms. At present, this is less
likely.
What you, Stan, have observed might be explainable as follows: you are
using a browser that can handle combining diacritics and also the
precomposed characters _but_ the glyphs for the latter are oddly designed
(consisting of a letter with the diacritic placed on the right of it).
If I tell IE to ignore fonts suggested on Web pages, then none of the
characters are correctly presented on most font choices. On Arial Unicode
MS, they are OK (technically - the glyphs are far too modern I'm afraid)
except Ṗ, which appears as mere P.
--
Yucca,
http://www.cs.tut.fi/~jkorpela/
Pages about Web authoring:
http://www.cs.tut.fi/~jkorpela/www.html 
