Re: Abstraction in HTML

Michael Johnson (michaelj@relay.relay.com)
Tue, 13 Dec 94 07:18:54 EST


All right, let's just dispose of this notion that merging different variations
on a list into a single tag is going to simplify anything. But before that,
let me make an observation: someone said that when they look at a list, they
see an assemblage of unordered items. What that person described was an
unordered list. If there is a list of items where the order matters, then they
should use an ordered list. The two are distinct conceptual elements.

Now, let's look at this "raising of abstraction level". As it stands now,
there are two tag names. One for ordered list, one for unordered list. We will
ignore DIR and MENU since they are deprecated.

If the two were merged into a single tag, then there would need to be a way to
differentiate them. At a minimum, you would then have one tag name, one
attribute name, and two attribute value tokens that a user writing HTML by
hand would have to remember. You have not simplified anything, in fact you've
made it harder to write HTML by hand. Where before a user had to remember two
identifiers, now they must remember four identifiers at a minimum. That is not
simplification.

As for making it easier to code a program to handle the two list types, the
program becomes more complicated too. Rather than having two cases in a switch
block to handle the two different lists (which makes it an atomic operation
to choose between the two) the program now must make the switch to the list
case, then must fetch the attribute which distinguishes the two list types,
then must parse the attribute value and figure out which type of list the
writer wanted. That is not simplification either, that's a significant
relative increase in processing time for something that ought to be simple.

I trust I've made my point.

Michael Johnson
Relay Technology, Inc.