Date: Fri, 26 Sep 1997 13:49:39 +0200 (MET DST) From: =?iso-8859-1?Q?Martin_J=2E_D=FCrst?= <firstname.lastname@example.org> To: Peter Flynn <email@example.com> cc: firstname.lastname@example.org In-Reply-To: <199709252055.VAA05006@imbolc.ucc.ie> Message-ID: <Pine.SUN.3.96.970926134250.361f-100000@enoshima> Subject: Re: ID and NAME On 25 Sep 1997, Peter Flynn wrote: > MZ writes: > I know ID and NAME occupy the same namespace. > > I thought the whole point about NAME was that it _didn't_ occupy > the same namespace as ID. Unless someone's changed it recently. > ID is declared as ID; NAME is CDATA last time I looked. > > <A NAME="fubar" ID="fubar">... > > Perfectly OK, precisely because they _don't_ occupy the same > namespace. If you want to be exact, their namespaces overlap, for historical reasons. It's impossible to declare ID as CDATA in SGML. However, NAME has quite some deployed use with values other than those very restricted ones allowed for ID. Also, ID is, from an SGML viewpoint, case-insensitive (i.e. FUBAR would be the same as fubar and FuBaR,...), where NAME traditionally and in all major implementations is case-sensitive (shorter code :-). This, together with the problems that case-sensitivity does not easily scale in an international setting (German sz, Turkish dotless i,...), let to the solution that: - The namespaces are merged (but you can have names in NAME that you can't in ID) - The names are case-sensitive (i.e. case matters) - It is not allowed/suggested to use names that only differ in case, because this might give bad surprises (an SGML validator interprets them as being the same, and so barks if one element is ID=fubar and the other is ID=FUBAR, but some other application may find it fine). That pretty much allows everybody to be happy :-). Regards, Martin.