Message-Id: <199802182151.GAA26669@sh.w3.mag.keio.ac.jp> Date: Wed, 18 Feb 1998 22:11:28 +0900 To: Ian Hickson <firstname.lastname@example.org>, email@example.com From: "Martin J. Duerst" <firstname.lastname@example.org> In-Reply-To: <000401bd3bee$0f40c480$cf20268a@hpxu> Subject: Re: Odd case sensitive rules regarding <A name="abc"> At 16:50 98/02/17 -0500, Ian Hickson wrote: > Can someone explan the logic in this? Yes. The NAME attribute started out in HTML to target a specific place in a page. The original implementations did not fold case (i.e. were case-sensitive, because that was easier to implement) and this has been continued. The ID attribute is used in SGML (where actually any attribute can be of type ID if declared correctly). For HTML, it became of interest when it was realised that not only the A element could be targeted; e.g. in stylesheets. The ID attribute in SGML is case-insensitive. Both attributes have a very similar function, and so it makes sense to have them use one single namespace. The rules were developped to cover all these cases. > >From the HTML4 spec:  > > > Section 12.2.1 Syntax of anchor names > > An anchor name is the value of either the name or id attribute > > when used in the context of anchors. Anchor names must observe > > the following rules: > > * Uniqueness: Anchor names must be unique within a document. > > Anchor names that differ only in case may not appear in the > > same document > > * String matching: Comparisons between fragment identifiers > > and anchor names must be done by exact (case-sensitive) match. > > So the following code is illegal: > > ---------------- > <P ID=ONE>... > <P ID=one>... > <P> <A HREF="ONE">Link to first paragraph</A> > <P> <A HREF="one">Link to second paragraph</A> > ---------------- Yes, because the two IDs have, in SGML, the same ID, and this is not allowed. > Yet the following code should do nothing: > ---------------- > <P ID=one>... > <P> <A HREF="ONE">Link to the paragraph</A> > ---------------- I guess it has to be <A HREF="#ONE">, but otherwise you are right. The whole thing looks illogical if you think only about machines: If a machine can distinguish between "one" and "ONE", why shoudn't both be allowed. If you also think about humans, and about the many languages there are on the planet, it can make more sense: For some people, distinguishing upper case and lower case is not so easy. On the other hand, various languages vary in how they associate upper-case and lower-case letters (e.g. in Turkish, an "i" is not the lower case of an "I"). So it makes more sense to have some "safe area" between what matches and what doesn't. Regards, Martin.