World: r3wp

my rewrite function works quite well for search and replace. it still 
has the limitations of parse, though, but they don't seem a huge 
problem so far.

example: assume the AST for 3*1+2+0 is [plus [plus [mult [lit 3] 
[lit 1]] [lit 2]] [lit 0]]

we want to compile it to a forth-like processor

>> rewrite [
[        plus [plus [mult [lit 3] [lit 1]] [lit 2]] [lit 0]
[    ] [
[        ['lit set x number!] [push (x)]
[        ['plus into ['lit 1 1 0] set x block!] [(x)]
[        ['plus set x block! into ['lit 1 1 0]] [(x)]
[        ['mult into ['lit 1 1 1] set x block!] [(x)]
[        ['mult set x block! into ['lit 1 1 1]] [(x)]
[        ['plus set x block! set y block!] [(x) (y) add]
[        ['mult set x block! set y block!] [(x) (y) mul]
[    ]
== [push 3 push 2 add
]

(rewrite is currently 25 lines of code)

do you have common examples that you consider problematic for parse? 
we can probably use rewrite on the parse rules themselves to extend 
them in a similar way that compile-rules does.

That's pretty cool. But, isn't the output of the rewrite above missing 
the multiply ?

Oh, I see, it has optimized the expression.

Well, I'm back after a long time off the air... Now back online and 
talking from a brand new iMac, and I enjoy it :-)

I submitted RebelXML to rebol.org quite some time from now. I can 
see it as been examined by some people. Well, it should be great 
if I could get some feedback about it.

We'll use it in production very soon. I took a SAX approach because 
we've to manipulate big XML files, and it was a mean to boost performences.

why rebel and not rebol?

Is REBOL not pronounce REBEL ? It was just for the fun...

implementing schema is sooo much fun!

I wonder if REBOL can get standards fully compliant XML parser/emitter 
....

(I hope my previous post was read as intended... ironically!)    
;-)  although schema is pretty well defined and actually not too 
badly designed, its just so huge.

Well, no reaction to RebelXML. Do I have to assume it has no bugs 
or nobody is using it ? :-))

Folks seem to be using/talking about it ... just do a google search 
on "RebelXML". ;)

Well, Ashley, this is very kind of you :-) But after checking the 
links, it seems Google isn't my friend anymore :-))

Seriously, I need this lib for my work. Is there anybody out there 
who has ANY comment about it ? Please ...

Well, I just had a quick look. It seems to be clean code, well organised. 
Perhaps I can find some time to test it out myself in a few days.

I do not know much about the greater issues of XML, however.

Has anybody done work on translating XML schemas (perhaps automagically) 
to REBOL parse rules that can work on REbelXML/RebXML like native 
syntax?

hehe... I have a validating schema parser which uses a modified RebXML 
engine... but I still have not had the balls to ask my client if 
we would be willing to make it open source.... I'm sure he would 
benefit from the extra time people might put on it.

I did not do it using parse for lack of knowledge ... and also the 
fact that schema and PARSE do not have the same traversal mechanisms... 
but its still very fast and error report is extremely precise (it 
gives you the full path of the error!)

the change in RebXML was to forego of a few limitiations and to allow 
direct xpath like useage of the loaded xml... which can't be done 
with RebXML out of the box.

What are people using to create XML ?

core

apple core or solaris core?

solaris , I am just emitting some simple rss feeds

The HTML Dialect can do a bit of XML

is this yours?

yes, posting some code to you privately.

Ok .. Ta

RebXML: http://home.tiscali.dk/john.niclasen/rebxml/

Good ... a few choices

RSP, I use it to generate RSS feeds from Cheyenne's blog.

REBOL's built in BUILD-MARKUP function can also be a good choice.

Which was the first apss at RSP (1 of 2 choices) I handede to Carl. 
I think build-markup does only support <%= %> not <% %> (which I 
always found a pity)

apss = pass

( took me a while to figure out that one ...  )

Just playing with rebelxml

>> set-xml-data/content 'header/to "[carl-:-rebol-:-com]"
== "<header><to>[carl-:-rebol-:-com]</to></header>"


Now if I start with "header" and "to", how would I get the same result?

<xs:complexType name="MessageType">
<xs:sequence>

<xs:element name="Header" minOccurs="1" maxOccurs="1" type="HeaderType" 
/>

<xs:element name="Body" minOccurs="1" maxOccurs="1" type="BodyType" 
/>
<xs:element name="Error">
<xs:complexType>
<xs:sequence>
<xs:element name="Code" minOccurs="1" maxOccurs="1">
<xs:simpleType>
<xs:restriction base="xs:string">
<xs:pattern value="600|601|602|900" />
</xs:restriction>
</xs:simpleType>
</xs:element>

<xs:element name="DescriptionCode" minOccurs="0" maxOccurs="1" type="an..3" 
/>

<xs:element name="Description" minOccurs="0" maxOccurs="1" type="an..70" 
/>
</xs:sequence>
</xs:complexType>
</xs:element>


If I have the schema looking like this ( no examples I'm afraid ), 
is something like this what I have to generate
<header> ... headertype .. </header>
<body>
<error>
<code> 600 </code>
<Descriptioncode />
<Description />
</error>
</body>

I probably need to wrap <message> tags around it too.

How are people creating large xml documents where there are large 
numbers of elements and where the data is being drawn from a database.

Model the document first as a Rebol object!, and then generate the 
xml from the object?

That's a good method.

Some databases have the ability to generate XML directly. That approach 
can be faster, depending on how much data the database has to process. 
It all depends on what database you are using though.

Has anyone used rebelxml.r ?

Pekr thinks there is a tool that converts xml to a rebol object ... 
anyone know what it is?

wait a bit, I'll find it - it was cool stuff from XML REBOL guru. 
That person used it for his work ...