Login | Register
My pages Projects Community openCollabNet

Discussions > dev > Fwd: dtds/xhtml, huh?

Discussion topic

Back to topic list

Fwd: dtds/xhtml, huh?


Author todd fahrner <tfahrner at collab dot net>
Full name todd fahrner <tfahrner at collab dot net>
Date 2002-06-07 19:24:08 PDT
Message Sorry this is out of order. This is the back story behind Pier's post.

Begin forwarded message:

On Fri, 7 Jun 2002, todd fahrner wrote:
> On Friday, June 7, 2002, at 12:35 , Brian Behlendorf wrote:
>> Is Pier smoking crack?
> Maybe. I assume his objection is to our using the term XHTML (in the
> formal public identifier (FPI) part of the DOCTYPE) when W3C uses the
> same string to identify its definition of same. There's plenty of other
> differentiating info there, though - most especially the fully-qualified
> URI of the DTD, so it's not like we're trying to commit fraud. But yeah,
> you're not supposed to stomp on another issuing organization's FPI.
> Neither ISO nor W3C have sent the cease-and-desist letter yet.
> I think the most kosher thing to do would be to use a system identifier
> instead of a (clashing) formal public one: <!DOCTYPE HTML SYSTEM
> "http://style.tigris.​org/nonav/tigris_tra​nsitional.dtd">.
> We don't do this for 2 reasons; both are more expedient than principled.
> Browsers and other processors of text/html (as opposed to
> application/sgml) tend to be really stupid and you have to stoop to
> their level sometimes to get things done. This is a devolutionary cycle,
> but still some hacks are less harmful than others.
> 1. We need browsers to think they're getting stuff made by "very picky
> CSS developers", so they'll behave somewhat more consistently with the
> specs, and hence each other. Mozilla, MacIE5, and WinIE6 (to date)
> implement a DOCTYPE-based behavioral switch. They look for a DOCTYPE in
> incoming text/html as a shibboleth. If the DOCTYPE passes whatever tests
> they apply to this string, they stop trying to emulate many of the bugs
> evident in previous versions of the product line, or in more popular
> products. More here:
> <http://diveintomark.​org/archives/2002/05​/29.html#quirks_mode​>. Anyway,
> our DOCTYPE has been composed to fool these browsers into thinking they
> are getting W3C XHTML 1.0, so they'll be on best behavior. Quick demo:
> in WinIE6, on any SourceCast screen, reload after editing/omitting our
> DOCTYPE. Conversely, try adding our DOCTYPE to most any screen of SC
> before 1.2, and view in Mozilla - it will likely fall apart. That's the
> extent to which we were depending on nonstandard behaviors that are
> necessarily not reliable across UAs.
> 2. Our DTD is mostly the same as W3C XHTML 1.0 Transitional, with most
> departures derived directly from the non-Transitional, "Strict" flavor
> of same. All of this is noted in the DTD prologue. To say that our
> markup "has NOTHING to do with XHTML" is tough to defend. Maybe Pier is
> offended that some of the same "results first" kind of thinking that has
> made HTML such a travesty seems to be lapping upon the shores of XML
> purity, where errors generally are not to be tolerated. I see it the
> other way around: we're trying to rescue what is useful in HTML by
> writing and presenting it in ways that will let even dumb HTML browsers
> parse it as a tree - the same one - so they can decorate it with CSS per
> spec: <http://www.w3.org/TR​/CSS2/intro.html#pro​cessing-model>, and we
> can have fast simple pages that look nice. That could be plain old HTML
> with a well-formedness constraint, but see #1.
> So why don't we just write proper W3C XHTML? Because that would
> proscribe support for too many popular effects in Netscape 4.x (like
> full-bleed banners, borderless images as submit buttons, text that wraps
> as expected in textarea, etc) and WinIE5 (many table formatting issues).
> Let me know when we can start supporting WinIE5 only about as well as we
> do Nav4 today). Besides, I think many of our DTD deltas from their W3C
> models are more supportive of other W3C recs than their own.
> I can't parse "they are NOT html, or at least that's how my browser
> thinks they are". Does Pier's funk start before or after he views
> source? <g>
>> Brian
>> ---------- Forwarded message ----------
>> Date: Fri, 07 Jun 2002 10:47:31 +0100
>> From: Pier Fumagalli <pier at betaversion dot org>
>> To: Brian Behlendorf <brian at collab dot net>, Stefano Mazzocchi
>> <stefano at apache dot org>
>> Cc: Steven Noels <stevenn@outertho​ught.org>,
>> Nicola Ken Barozzi <nicolaken at apache dot org>, Sam Ruby
>> <rubys at us dot ibm dot com>,
>> root at apache dot org
>> Subject: Re: [HEADS UP] Forrest is running :-)
>> Brian Behlendorf <brian at collab dot net> wrote:
>>> On Thu, 6 Jun 2002, Stefano Mazzocchi wrote:
>>>> But I'm pretty sure you'll like what we do when you see the changes
>>>> in
>>>> load sucked by xml.apache.org when we bring the pages down from a
>>>> 100Kb
>>>> to a few Kb and from 50 req/page to 3 req/page.
>>>> For now, you can check both http://xml.apache.org/cocoon and
>>>> http://xml.apache.org/forrest which use different skins but where
>>>> both
>>>> optimized for load, yet without sacrificing usability and visual
>>>> appeal.
>>> Along these lines, you should check out the style.tigris.org project -
>>> it's the baseline stylesheets we're using for Scarab (and much of
>>> sourcecast), and they've had a lot of attention from some very picky
>>> CSS
>>> developers.
>> The only problem is that they are NOT html, or at least that's how my
>> browser thinks they are:
>> <!DOCTYPE html PUBLIC "-//CollabNet//DTD XHTML 1.0 Transitional//EN"
>> "http://www.collab.ne​t/dtds/collabnet_tra​nsitional_10.dtd">
>> This doctype is WRONG, it indicates a kind of document which has
>> do with XHTML...
>> XML is not a hack!!! :) :) :)
>> Pier

To unsubscribe, e-mail: dev-unsubscribe@styl​e.tigris.org
For additional commands, e-mail: dev-help at style dot tigris dot org

« Previous message in topic | 1 of 2 | Next message in topic »


Show all messages in topic

Fwd: dtds/xhtml, huh? todd fahrner <tfahrner at collab dot net> todd fahrner <tfahrner at collab dot net> 2002-06-07 19:24:08 PDT
     Re: dtds/xhtml, huh? ianosh Pier Fumagalli 2002-06-10 04:57:09 PDT
Messages per page: