Lists Home |
Date Index |
[Bill de hÓra <firstname.lastname@example.org> wrote:]
> Granted we've been here before about how developers find state machines
> awkward but it does leave open the possibility of being declared and
> then autogeneratated. Was this approach never taken with SGML?
>>Sure, though never in any standardized fashion. The frustrations with SAX
>>are recapitulations of the same frustrations with the output of (n)sgmls -
>>no surprise, since SAX is modeled on that implementation of ESIS, and even
>>more closely on David Megginson's SGMLS.pm package. (Which is to say, SAX
>>was not in any way an advance in the state of the art.)
SAX and ESIS are fundamentally different in one way - ESIS is a *notation*
produced by NSGMLS.
SAX is an API. You cannot grep an API. Hence PYX.
The fact that SAX (an API) flourised and PYX (a notation) withered
on the vine could be seen as a bellweather here.
The majority of programmers *prefer* APIs to notations. Therefore
SOAP will be a runaway success.
<Bart>Dow! I better start liking APIs before the market gives
me a Wedgie for stepping out of line.</Bart>
 Strictly speaking, ESIS is an abstract description of an infoset. The
of NSGMLS is a reification of that abstraction into a syntax. James Clark
invented the syntax - "(" for open element "-" for data, "A" for attribute etc.