Re: Notes on the process

Gavin Nicol:

> [...] Let's take Bill's example to the 
> limit, and say that the obboard computer sent out:
> 
>   <ACCIDENT>
>   <VEHICLE>sd;ghligk;db/lflgniupg nbpppv</VEHICLE>
>   <LATITUDE>klnsfvouvjisbdfubfibdbufubufbf/LATITUDE>
>   <`340-968gn]0984588y873**(&&E$*(&& m8qf9hnc 
>   </ACCIcore dumped, your fault!
> 
> would you still expect it to work? What about if it sends out the
> wrong vehicle identification, and the wrong positional data? Suddenly
> you have an ER team rushing out to the wrong place, looking for
> the wrong person.
>
> For the situation Bill outlined, I would expect a simple radio
> transmitter to work far more effectively: you can use triangulation
> to locate the source, and encode the radio signals to include a
> great deal of data redundancy in a much smaller number of bits.

I can't predict that someone won't devise an effective internet-based system
that integrates numerous technologies, including XML, to let me report that
"I've fallen, and I can't get up!" And so what? Any accident reporting system
can fail; there are no 100% guarantees. Each of us has seen astonishing
freeway pile-ups that could disable a vehicle's accident reporting apparatus.
Did the accident cause incorrect XML or a radio transmitter failure?

I think that these scenario discussions are interesting but ultimately they
involve other ideas:

(1) XML's problem space: what kinds of applications is it designed to support?

(2) Fault-tolerance: what should happen to data when, for any reason, it is
not well formed? (And doesn't that relate to the kind of application being
considered?)

(3) Imagination: just what kinds of limits can anyone put on the future of
Internet-based applications? The marketplace always surprises me.

Maybe none of these questions is truly relevant to the ERB's work, even if
they may be interesting. These questions certainly are popping up lately;
should they? Maybe Henry Thompson really does have it right:

> Let's make the core and link (and style) language specs just that:
> Descriptions of the grammar of valid and well-formed XML documents.
> No mention of processing, error policy, behaviour, etc.  Semantics,
> when it must be referred to, would be in terms of abstract structures
> and relations, NOT 'the processor'.

Received on Friday, 9 May 1997 11:15:42 UTC