Re: XML tools for Octave

[Bill Denney]

On Thu, 29 Jun 2006, Andy Adler wrote:

> But the issue here is the extensibility at the heart of XML. The point 
> is that you should be able to add extra information (the <metadata/>) 
> and a parser for the old format should still work

That sounds good, and I like what you have below with a small change to 
how ORDER is used.

> Here is my latest idea for the format:
> <a b="c" d="e"> text <f g="h"/> more <é->text</é-> <f>data</f></a>
>
> v.a.UTFNAME         : "a"
> v.a.NAMESPACE       : ""
> v.a.ORDER           : 1
> v.a.ATTS.b          : "c"
> v.a.ATTS.d          : "e"
> v.a.TEXT{1}         : " text "
> v.a.TEXT{2}         : " more "
> v.a.TEXT{3}         : " "
>
> v.a.f{1}.UTFNAME    : "f"
> v.a.f{1}.NAMESPACE  : ""
> v.a.f{1}.ORDER      : 2
> v.a.f{1}.ATTS.g     : "h"
>
> v.a.f{2}.UTFNAME    : "f"
> v.a.f{2}.NAMESPACE  : ""
> v.a.f{2}.ORDER      : 6
> v.a.f{2}.TEXT{1}    : "data"
>
> v.a.e_{1}.UTFNAME    : "é-"
> v.a.e_{1}.NAMESPACE  : ""
> v.a.e_{1}.ORDER      : 4
> v.a.e_{1}.TEXT{1}    : "text"
>
> Here I keep the order of the tags in each tag. The text elements are
> then interspersed between them. Names (like é- are modified, but
> the original is kept, as required).
>
> Comments?

If you could count the text elements when keeping track of the order 
variable, then I think that everything would work the way I want because 
then you would have reversibility.  The way that ORDER could give 
reversibility is if they were specified with counting the text but not 
specifically numbering it i.e.

v.a.ORDER           : 1
v.a.f{1}.ORDER      : 3
v.a.f{2}.ORDER      : 8
v.a.e_{1}.ORDER     : 5

with the rest of the data the same.  That way, someone who cares about 
reversing it can put in a text element when order is missing an element. 
I would like that parser.  It is fully defined because you must fully 
handle the e_ element before you can leave so you can know that its TEXT 
element must be 6 and that the third text must be 7.  If you wanted to 
make reversing parsing simpler, you could add a TEXTORDER field (a vector 
with each element corresponding to one element of the cell array TEXT) so 
that the book keeping would all be done during the original parse.

Bill

--
"The chaos escalated when a group of armed revolutionaries from the
Honduran countryside raided JCPenney."
  -- Mall Of Central America Looted On Opening Day, The Onion