Cart
[Home] [By Thread] [By Date] [Recent Entries]
From: "Tim Bray" <tbray@t...> > > Check out http://tbray.org/tag/utf-8+names.html Instead of throwing an error for a missing semicolon, why not let that case fall under section 4. Since the set of pre-defined reference names are known, you also know the maximum length possible. If you reach length+1 characters, you know it's not a reference and leave it as-is. It also means that a lone ampersand won't be an error. I suspect the motivation to require the semicolon is for consistency with XML's same requirements. However, this disallows some valid UTF-8 from being valid UTF-8+names as well. While this may be fine if used solely within the context of XML (which would throw an error as soon as it hit the invalid reference anyhow), I suspect people will try using this outside of XML as well. Also, I suspect that using the same format as XML/SGML's references is going to confuse people. Maybe use a similary format such as #name; or @name;. This way, at least the two references (UTF-8+name and XML) would not be confused. --- Seairth Jacobs (seairth@s...) Looking: http://www.seairth.com/blog/65
|
