parallel lineal numerical string comparison algorithm

[Ruslan Zasukhin]

on 5/18/03 8:09 AM, Mark Brownell at gizmotron at earthlink.net wrote:

> Hi, Ruslan, Andy,
> 
> A few years ago I was toying with the idea of pull-parser concepts for
> XML. I found great success using a parallel lineal numerical string
> comparison algorithm; the original source code is still referenced from
> the download page for Director V4MD at the paradigmasoft.com page.  I
> have been using a combination of a database table containing only one
> record per row. Each record is in fact an object containing a small
> section of a fuller XML document.
> 
> Example Object:
> <object>
>  <transactionNumber>1234567 </transactionNumber>
>  <name>Bob T. Customer</name>
>  <product name="Super Fishing Real XX10" code="121212-34"</product>
> </object>
> 
> A pull-parser acts on the object by isolating the different elements
> and attributes and manipulating them outside of the SQL call that is
> used just to access each object. In this way it is possible to create
> an entire text based database based on the XML found in each object.
> Although this is a hybrid solution that is still dependent on the SQL
> call, to get each object, the real power is the pull-parser and its
> ease of use in text or XML transformation projects. I've used the PNLP
> technique in Director, RB, and Runtime Revolution.
> 
> So a high speed PNLP that could pull attributes and elements from XML
> objects located inside of  tables could be used to create an easy to
> use Query Language now. (think string comparison)
> 
> I don't know if this is on topic but I know it is easy to access data
> and do different interface things with that data if I need to using
> this technique.

Yes, if to speak about idea of Andy, then yes, such algorithm can be useful.

Mark, I do not understand although one thing.
So your table has only one field, yes?
I assume TEXT field.

So you store in table for example 100,000 records.
Each record is one XML block that contains several attributes.
Correct?

Now you on some rule extract one of record, and use your PNLP algorithm to
extract attributes from this ONE RECORD.

But Mark, I believe it will be MUCH MORE effective if you parse XML block on
attributes BEFORE put it into database, and then store only VALUES of this
attributes into Table with N fields. This is more effective because now
1) db do not keep a lots of redundant names of attributes in each record.
2) now you can use indexes of Valentina
3) numeric values stored as numbers but not as text.

What you think about this ?

May be each of your record, have DIFFERENT set of attributes?
But then you still can build correct Relational tables to store such
records. 


-- 
Best regards,
Ruslan Zasukhin      [ I feel the need...the need for speed ]
-------------------------------------------------------------
e-mail: ruslan at paradigmasoft.com
web: http://www.paradigmasoft.com

To subscribe to the Valentina mail list go to:
http://lists.macserve.net/mailman/listinfo/valentina
-------------------------------------------------------------