Re: Is there a way...

On Mon, 16 Apr 2001 08:53:54 -0700, "Chris Wilson"
<cwilso@microsoft.com> wrote:

>Jan Roland Eriksson [mailto:jrexon@newsguy.com] wrote:
>>What will the upcoming MSIE6 do with e.g. the following?
>>  <!DOCTYPE HTML PUBLIC "+//IDN htmlc.cx//DTD RcM1.0//EN" [
>>    <?ArcBase RcM>
>>    ...lots of other architectural and notation stuff here...
>>  >]
>>  <HTML>
>>     ...
>>  </HTML>

>I presume you actually meant to put "]>" in line six, not ">]".

Yes, a definite typo on my part.

>>This is a basic skeleton of what is known in SGML as an
>>"architecture processing instruction". You can stay friendly
>>by saying that MSIE6 will assume "full CSS compliant rendering
>>mode", if not? well DocZilla and Opera are already usable to
>>present documents with an "odd" prologue.

>IE 6 will indeed assume CSS compliant mode.

That's a good start...

>However, it gets confused by what it sees as nested angle
>brackets, and terminates the PI after "RcM>".

Would that behavior really be necessary for a document that has been
served as text/html, and where the <!DOCTYPE... declaration has already
been found to be outside of "W3C standard recognition" ?

>Note that we've never claimed to a full SGML processing engine
>in IE, and it's highly unlikely that we ever will - as XML has largely
>replaced SGML as a syntax language for markup schema.

I think that what I'm asking for at this point is not SGML compliance,
but rather the best possible SGML ignorance, that I can get from
browsers that already has decided to use <!DOCTYPE... recognition for a
purpose it was never intended for in the first place.

As I see it today (when it seems just inevitable that I must accept
doctype sniffing as a "fact of life") is that any browser that gets
served a 'Content-Type: text/html' doc, that also sports an
"unrecognizable" <!DOCTYPE... declaration, should be able to look only
at that <!DOCTYPE... and a possible System Identifier to just send the
message to its rendering engine that this coming thing is to be rendered
"strict".

Following that already made decision, browsers should also be fully
capable of "ignorance" visavi whatever it finds between the <!DOCTYPE...
thingy and the first occurrence of (or a correctly implied) <HTML...>
tag.

There is no need for text/html docs to be read further for PI's; just
find the first <HTML...> tag and if the <!DOCTYPE... was "unknown" (as
per W3 recs) go on to render the doc in strict mode. In that case at
least I will be happy. I mean "tag-soup" is already "tag-soup", right?
I just want a way to get at that strict renderer for sure.

That prologue I "illustrated" up there is basically only relevant to
another parsing/processing event where nsgmls/SGMLNORM plus a specific
application is what will know what to do with that "intermediate stuff".

I for one have reasons to ask for a permission to include the arch form
processing machinery where it belongs, for very good reasons inline with
my own daily work. And no; XML does not cut it where I am, sorry about
that, so I still need a decent and "SGML tolerant" HTML document viewer.
(do I have to represent a multi million company, or be a member of "the
inner crowd" just to ask for that?)

>>>...the CSS Working Group thought it would be a good idea
>>>to correct the spec...

>>They did not "correct it",
>>they "fucked up" and that's all there is to it.

>You should really moderate your language if you wish to be taken
>seriously.

They had the options available to do what they did, or to create a
really solid CSS2.0.1 recommendation. IMO they did take the wrong turn
on that. (is that an acceptable level of moderation?)

The whole "kit and kabodle" is something that I argued for already in
August 99...  Message-ID: <37ad1440.671691491@enews.newsguy.com>

And there is also a "Hen and Egg" problem behind this. I have ended up
being "bashed around" at times, after I have spent numbers of evenings
of my own spare time to come up with suggestions on how to implement a
rendering mode selection method that might be acceptable to all.
Comments coming back has been mostly "out of ridicule" and has also not
showed much of "analysis of suggestion" or "new ideas triggered" either.

I was at one time (I think?) "closer than one inch" to convince DB that
an "over riding HTTP solution" could be made to "work for all"
_together_ with <!DOCTYPE... sniffing.

All that remains from that today is a Bugzilla entry that has been
tossed back and forth a couple of times between the protocol and parsing
groups, where no one anymore wants to touch the idea, not even with
pliers and safety gloves.

So who is the Hen and who is the Egg? Me and/or W3C?

-- 
Roland
[force a dog into a corner, and he will bite to regain freedom]

Received on Monday, 16 April 2001 20:30:25 UTC