RE: A simple guy with a simple problem

[Staffan Måhlén <staffan.mahlen@ausys.se>]

Being a bit of a Bob incarnation (no association with a small town called
Twin Peaks) i wonder whether my problem had any relevance to what Sean
McGrath is saying about layering, albeit from a very different point of
view. From Richard Tobins very illuminating post [1] i believe i gathered
the following.

Given:
The recommendation had to chose between two unattractive choices, and
avoided the (theoretically) worst one.

"Most" parsers are "incompliant" with regards to a simple theoretically
wellformed document (which is useless and bad practice) and thus also
reintroduce(!?) the problem that the recommendation strives to avoid.

Makes me believe:
Removing internal subsets would help simplify the problem and would make it
possible to make PE declarations mandatory, making parsers compliant in that
respect... 

At a wild guess that change would be pretty much impossible. So how about
just making PE declared a WFC and make the best practice a hard rule in the
rec as well as in reference parsers? Then again, best practice (and common
sense i suppose) already fix this so i guess it would really just be for
simplified comprehension. Better (and vastly more initiated) minds than mine
has apparantly been messing with this one, so if there are any other reasons
for this "mismatch" or if i still haven't got it right, i would be
interested in knowing. I am not quite convinced the items in the second
edition errata regarding this are "best practice".

Thanks for the help,
  Staffan Måhlén
 
[1] http://lists.xml.org/archives/xml-dev/200103/msg00626.html