But I think that's merely a reflection of reality: not all browsers
handle caching/history the same way, and information providers must
not count on any particular behaviour in this arena.
-------------------------------------------------------------------
I strongly disaggree. One of the intents of the BASE URL construct is so
that authors may create documents with relative URLs and know exactly how
they are going to behave.
If browsers are not consistent, in their behaviour, then I simply can not
count on ever using <a href = "#Chapter_1"> in a HTML document, and know
that it will work (it always will in Mosaic, it may or may not in Netscape).
This is not acceptable. We either need to specify a consistent behaviour
for stand-alone fragment identifiers, or invent a new tag to guarentee
this behaviour. Say, something like <a href="section:Chapter_1">?
I can't fathom any case where a stand-alone fragment identifier should
refer to anything but the currently-loaded instance of a document. If
someone can show me such a case, I'll back off, but it's just so
counter-intuitive that I can't understand why it should be any other way.