Lists Home |
Date Index |
First, I was not arguing a specific subset. I don't even have a feel for
all of what would be left out, though there are certainly a few bits that I
can personally do without. All I am arguing is that I am seeing the same
sort of resistance to an XML subset as I understand there was to initial
conversations about an SGML subset. XML was not designed to replace SGML,
just to handle a certain domain of problems better.
Now that we have a few years experience with XML under our collective belt,
people are making the same (general) arguments about XML. Even if such a
subset was created, I realize that there would be dissatisfied users of that
set as well. That's just the way life is. But this should not be a
deterrent from finding a subset if it is going to do a better job than XML
as solving a certain domain of problems.
Having said that...
From: "Wayne Steele" <firstname.lastname@example.org>
> 1. You say "Lets abolish mixed content." This makes anyone who uses XML
> document markup plotz. Are you saying Docbook, XHTML, and FO are not
> in your subset?
Supposing mixed content was left out, then yes, those that depended on it
would not be usable by processors that only supported the subset. This
would be no different than an XML processor not being able to handle some
SGML specifications. On the other hand, such parsers may be more optimized,
schema languages may be designed different, etc. There's always a
> 2. PIs are usually first against the wall in most people's subsets (see
> SOAP). I noticed you see them as useful.
I don't have an opinion on them one way or another.
> 3. Dropping the XML Declaration ( the <?xml ... ?> thingy ) has
> side effects. This is how you declare the character encoding of your
> document. Are you suggesting that all XML documents (in your subset)
> be in either UTF-16 or UTF-8? Will the people you hope to work with accept
> this - or have you just alienated most east asians?
I'm not suggesting anything. It may be appropriate to use a different PI
instead. On the other hand, it may be appropriate to use yet another PI.
For instance, you might have something like:
<?xml version="1.0"?><?subsetname version="1.0"?>
(note: this is just a thought off the top of my head. Like everything else
I say here, this is just speculation and not necessarily of preference of
> 4. The few things that everyone pretty much agrees can be dropped, usually
> are using the "parse and ignore" method. When was the last time you saw
> anyone paying attention to Notations or Attribute declarations (other than
> ID & IDREF).
Yes, but they must still be parsed even if they are ignored. That wastes
time and resources. Some may consider this a weak argument, but it is still
a legitimate one.
Again, I am not advocating any specific subset of XML here. I am just
pointing out the "shoe on the other foot" attitude (on both sides) that I
keep seeing here...