This is conforming behaviour, unless the document is somehow
labelled as having a document character set with code positions
above 255.
> I believe this behavior should be discouraged; they should be doing
> something like the following on a Latin1 platform:
>
> int ch = atoi ( numCharBuf )
> if ( ch > 255 || ! isprint(ch) )
> ch = <your favorite substitution character code that produces a box>;
We can't go back in time and change the installed base, so this
is largely a moot point. But I don't want to encourage the "it
works in mosaic" syndrome. Those browsers _should_ report an
error:
Bad numeric character reference `ř' at line 27.
or some such, so that the author has incentive to fix the broken document.
An HTML 2.0 user agent must be 8-bit clean, and that's it. Support for
"wide characters" and such is essentially out of scope for 2.0.
Let's get busy on the I18N document!
Dan