From mboxrd@z Thu Jan  1 00:00:00 1970
Message-ID: <fbb4f5bebcf5822a894afceb2466f92d@vitanuova.com>
To: 9fans@cse.psu.edu
Subject: Re: [9fans] OT: small xml parser found!
From: rog@vitanuova.com
In-Reply-To: <1077206087.1790.18.camel@pc118>
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
Date: Thu, 19 Feb 2004 17:15:44 +0000
Topicbox-Message-UUID: eedc128e-eacc-11e9-9e20-41e7f4b1d025

> >> Why isn't Expat suitable?
> >
> > callbacks are horrible.
>
> Reasons?.
>
> Callback-(handler)s are interrupt-(handler)s
> [...]

can i just add this reference:

http://www.codepedia.com/wiki/display.aspx?WikiID=1&pagename=thunks

as an example of a really nasty way (note the machine dependent bit)
of getting around some of the limitations of a callback-based system.

if i was going to use expat, i'd wrap it up with the plan 9 threads
library, so that constructs would arrive on a channel, then at
least things would be marginally more bearable.

but even then, you'd probably want to put a function call interface
around the channel, as otherwise it's really quite awkward
skipping subtrees you don't want to know about.

i wasn't too dissatisfied with the xml(2) interface in inferno,
which looks something like (limbo syntax, i'm afraid):

	open: fn(fd: ref Sys->FD): ref Parser;
	Parser: adt {
		next: fn(p: self ref Parser): ref Item;
		down: fn(p: self ref Parser);
		up: fn(p: self ref Parser);
	};

next() gives you the next item at the current level (nil if we're
at the end of the enclosing block); down() delves into the most recently
returned element, and up() ascends a level.

one advantage of doing it this way is that potentially the parser
can drive how the parsing takes place; the above interface
takes advantage of that by allowing random access into
the XML (you can go back to a place you've previously marked).