[rdfweb-dev] Re: bio:olb --- text or hypertext?
wkearney99 at hotmail.com
Tue Aug 5 23:16:03 UTC 2003
Get out your whip because you're beating a dead horse here. In fact, you're
beating the bones of the dead horse in the remains of a barn long burned down.
While you can wish all you want about demanding fully specified content within
elements. You're not going to get it. Perhaps not even in your lifetime.
What about the environment that *deliberately* wants to interchange tag soup?
To follow your dictate you'd expect the content to be massaged into something
proper and then what? What if the target system CAN'T consume your wonderously
well-formed data? Your conversion process would break an otherwise working
solution. Yeah, that's progress.
Look, I'm all for well-formed markup. But the *realities* make that neither
possible nor feasible at this point in time. Yes, most certainly, things should
be /strongly/ encouraged to emit natively well-formed data. But there's no
sense in forcing your dictates on them when the specs allow for an software
depends upon being able to use encoded markup.
But hey, if you're so hell-bent on eliminating it, take a hard look at what
tools are producing content that so insults your sensibilities. Write patches
that utterly eliminate the mistakes those programs are making. We'll wait for
Meanwhile, we'll encode.
----- Original Message -----
From: "Norman Walsh" <ndw at nwalsh.com>
To: "Bill Kearney" <wkearney99 at hotmail.com>
Cc: "rdfweb-dev" <rdfweb-dev at vapours.rdfweb.org>
Sent: Tuesday, August 05, 2003 2:59 PM
Subject: Re: bio:olb --- text or hypertext?
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> / "Bill Kearney" <wkearney99 at hotmail.com> was heard to say:
> |> | Regardless of whether it's right or not, we're trying to transport
> |> | user entered text embedded in XML. Whenever we do that we face this
> |> | issue. As developer's we get to choose whether to strip tags, but even
> |> | that can get exceedingly hard some times, as in Dan's example.
> |> I understand. But I think we should set user expections that elements
> |> are either text (no markup!) or markup (use XML!) and not this odd
> |> mixture of both where some layers see it as text and others see it as
> |> markup.
> | The realities are that tools prepping content are not capable of effectively
> | guaranteeing that the data inserted into the contents of an element are
> | constructed to be well-formed and valid XML. As a result compromises have
> | made that allow effective use of 99.9% of XML's goodness. That that .1%
> I respectfully disagree. If users come to expect that
> <someWrapper><![CDATA[try this:<br><a href="http://...">click
> is going to produce the visual effect of two lines of text and one
> link, then 99.9% of the XML goodness is obliterated. We'll be back to
> tag soup. Only worse, because now the tag soup will be nested and
> disguised. (I can't wait for the application that wants to nest this
> nested markup inside nested markup.)
> There are absolutely no circumstances in which escaped markup should
> ever be rendered. It is an unacceptable design.
> | (figuratively speaking) is the hard part and, for a great many applications,
> | entirely unnecessary. Yes, it would be 'better' if they got it right. A
> | things in life could be better... if only...
> Lots of things would be better. I'm all for pragmatic, 80/20
> solutions. But the corollary is important too: Some things are
> absolutely, fundamentally, irrevocably harmful if they're not done
> well enough.
More information about the foaf-dev