[Translet-general] RE: XPath To XML Translet

[Karthikeyan M. <mak...@ya...>]

--- Mani Subramani <msu...@li...> wrote:
> Hi Karthik, I am Mani, I am the one who give Pakshi  a hard time in
> ping pong :-) just kidding
>  

Good to hear that you are chasing him hard. He needs a lot of running
now!

> Thanks for introducing to your project. I read the description, its
> sounds very interesting. I have a few questions , please reply
> whenever you are free
>  
> 1. Do you always assume that you are going back from XSLT to XML ? In
> other words, can this tool be also used in a place where the 'view'
> path doesn't have XSLT , but for some reason there's a need to
> transform 'from-HTML Client-to-server' data back to XML

The short answer is, yes you can. The concept behind is actually much
simple. Once you want to create an XML document, you need to know it's
structure. If you know it's structure and have the complete XML
document in your mind (hypothetical at this time, as the document is
not yet created), you can create a mapping between each value in the
document to it's XPath in the hypothetical document. Unlike the
document itself which is hypothetical at this point, the mapping is
real. So, all you need to do is the reverse of what XSLT or XPointer
does. Parse the XPath and create the nodes on the way.

> 2. Can you describe a more concrete use case / example of the purpose
> of the tool.

Actually, this concept is already at work here in Vitalect, but, in a
much restricted and nascent form. The use case is that we have to
create XML documents from the form input user has submitted using HTML
form. There are many such form all over the application and we need to
create many different kinds of XML documents. I implemented this
translet concept for that. We have a mapping of the form field names to
it's actual XPath in the document. Once the form is submitted, we
retrieve the XPath value for each of the form field and create the
node/branch in the XML document.

> 3. How is this problem currently solved ? Is it by implementing XML
> 'writers' that use the DTD/Schema and data and make the XMLs ?
>  

From my own research on this, I found people mostly do it the hard way
(hard compared to the technique this translet uses). They write java
(c++ or whatever language) code to create each node and attach it to
it's parents etc., This may be the very obvious method, but one that
involves constant maintenance and one that is very less re-usable. Each
time you come-up with a new XML document or change your DTD, you have
rework the entire code.

> Thanks and good luck for the project, it sounds really interesting
>  

Thanks for your queries. Will keep you posted. Would welcome any
suggestion and ideas, I guess that is how an open source project grows.

> ~M
>  
-karthik

__________________________________________________
Do You Yahoo!?
Yahoo! Health - Feel better, live better
http://health.yahoo.com