Re: HTML 2.0 LAST CALL: Numeric character refs

Murray Maloney (murray@sco.COM)
Sat, 3 Jun 95 08:23:48 EDT

David - Morris writes:
> On Fri, 2 Jun 1995, Terry Allen wrote:
> Terry quoting Dan:
> > | Perhaps this is more clear:
> > |
> > | |On the other hand, references to undeclared entities and undefined
> > | |numeric character references (i.e. references to code positions that
> > | |are not in the domain of the document character set) should be treated
> > | |as data characters.
> >
> > | I'll take it out if Mr. Morris will stipulate.
> >
> > No. Mr Morris is not speaking for the WG on this point, and you had
> > consensus (if only by silence, but after much discussion) with the
> > language you had. I'll say it again, you had no business making this
> > change. I see I now have support from Martin, and I renew my
> > opposition to this last-minute change.

I have to agree with Terry on this. This is a last minute change
and it is does not conform with SGML in practice or theory.
> Actually, I can make the same silence argument re. WG concensus.
> During the lengthly 10646 dicussion, I raised the error issue
> several times and as I recall was generally not responded to. Dan
> eventually challenged me to provide words.

The difference being silence about something that was written
in the spec versus something that was being proposed on the
mailing list. Being in the spec is an indication that rough
concensus exists (or existed at one time). Raising an issue
on the mailing list only means that you have a mail account.
> I provided words when I had a stable reference point and Dan has
> done an elequoent job defending his change in the context of SGML
> conformance. (I prefer the revised wording at the top recently
> suggested as well).
> When 10646 or whatever is discussed again, this change is a
> sufficient solution and I would argue superior to an unknown
> character ICON which would require more user effort to resolve
> than simply looking up an integer in a book.

As a user interface solution to a problem, I would have to agree
with you that it is better to know the value of a character
that is "out of range" than to see an icon. But that is not
the issue, as we are not responsible for a user agent spec.
If you want to initiate a user agent spec, and specify the
rules for error handling, please feel free to proceed. I will
be happy to offer whathever help I can to make that spec
a useful and usable document.

But, we are talking about HTML, an application of SGML. Whether
we like it or not -- and I know that some don't -- the spec
must conform with SGML in every respect or it might as well
not bother at all.

Neither Dave nor Dan have raised any argument which justifies
breaking SGML conformance, and Terry has clearly demonstrated
and explained that the proposed changes do break conformance.
> What I will stipulate is that I won't raise this issue in response
> to the 'public' last call on the ietf list or whatever (doesn't
> mean I won't respond if an 'outsider' complains). So if
> Dan, Eric, and Tim or whatever group decides that the
> rough concensus is to remove my suggested change, so be it.

OK. So, it's time to send mail and be counted. Now you
know -- again -- where I stand.
Murray C. Maloney Internet:
Technical Publications Writer/Architect Uucp: ...uunet!sco!murray
SCO Canada, Inc. My Phone: (416) 960-4031
130 Bloor Street West, 10th Floor Fax: (416) 922-2704
Toronto, Ontario, Canada M5S 1N5 SCO Phone: (416) 922-1937
Disclaimer: I'm speaking for myself. 'T ain't nobody else to blame but me.
Sponsor member of Davenport Group Member of IETF HTML Working Group