Re: Final words, I think, on error handling

At 11:41 PM 05/06/97 -0400, Paul Prescod wrote:
>No, there will be new bad old days with broken links, mismatched
>attributes, invalid content models, broken stylesheets etc. etc. etc. I
>really don't understand how being draconian in this one case solves
>anything. 

Those are part of the *current* bad days, I think. The fact that we can't
bring on the millenium all at once doesn't mean we shouldn't improve the
world where we can.

>... For every improperly
>formed HTML document there is one that uses an element in an invalid
>place or an HTML element that has never been defined in a DTD anywhere

If this is true, then we get to eliminate 50% of the problems in one shot.
I didn't have any hopes *nearly* that high.

One other item in response to the current discussion, from a highly
reluctant "draconian" (I only apply the label at all since despite arguing
vociferously against that position, in the end I voted for it): 

There is another reason that two big warring companies can't just "agree" on
this. We all know that some company somewhere could claim a marketing
advantage by claiming to do cool error recovery in browsers. This would be
to their individual advantage, at great cost to the community (e.g., their
recovery won't be identical to others' and so we're unavoidably back to
having non-portable documents). Each company therefore must trust the other,
AND all future competitors, not to try to gain the upper hand by cheating. 

This is the tragedy of the commons: each shepherd must trust the others not
to try to gain the upper hand by grazing sheep on the commons. That's one
reason we end up with laws. And the XML spec is the applicable "law". It has
no truly binding force, but it provides a way to censure applications that
act against the common good. No company can do that, it has to be a
legitimate standards organization of some kind.

Put more simply, NS and MS *need* the force of our legislative weight to
help them achieve detente, not only with each other but with anyone else. At
least that's how it looks to me. If we give NS and MS grounds to refuse to
accept non-WF documents, I don't think it unreasonable to expect that there
will very soon be a lot fewer non-WF documents floating around.

I *like* error recovery, but I think it's worth sacrificing in browsers (or
"read-only" applications as Tim put it) for the common good of
interoperability. Note that it's still entirely welcome in editors, because
*there* it advances the common good by helping people produce WF documents
out the end. I also presume "editor" includes an editor-ish program that
transforms incoming cat fur into WF XML output.

Steve

Steven J. DeRose, Ph.D.
Chief Scientist
Inso Electronic Publishing Solutions
   (formerly EBT)

Received on Thursday, 8 May 1997 10:28:55 UTC