RE: report writer for XML

Your "guess" about what is a traditional control break report 
is accurate.  There are typically several levels of
control breaks, including one global (grand total) level.
In addition to totals and subtotals other measures are often 
desired as well, e.g. average, maximum, etc.  
In addition to the logical breaks when a value changes
reports typically also have page headers and footers,
as you mention.

A different but also important style is a "cross tabs" report.
Here both horizontal and vertical axes are used to constrain
the contents of the cells.

Yes, ideally the data is sorted appropriately, and then the 
report can be produced efficiently in one pass.
 
Hopefully helpfully yours,
Steve
-- 
Steven Tolkin          steve.tolkin@fmr.com      617-563-0516 
Fidelity Investments   82 Devonshire St. V1D     Boston MA 02109
There is nothing so practical as a good theory.  Comments are by me, 
not Fidelity Investments, its subsidiaries or affiliates.


> -----Original Message-----
> From: Per Bothner [mailto:per@bothner.com]
> Sent: Thursday, January 24, 2002 11:17 AM
> To: Tolkin, Steve
> Cc: www-ql@w3.org
> Subject: Re: report writer for XML
> 
> 
> Tolkin, Steve wrote:
> > We are thinking about storing a collection of XML instance 
> documents,
> > that will all conform to a particular DTD.  But we do not understand
> > how to run traditional control break reports on these documents.
> 
> Do you have a short summary of what you mean by "traditional control
> break reports"?  If so, it might be useful to post that.
> 
> Way back I had a little experience in the field of "report generation"
> (when I was a Cobol progremmer).  My guess as to what you might mean,
> in an xml context:  You have a long list of elements of the same basic
> type.  You want to group elements that have the same value for some
> attribute or child element, and after processing those elements, add
> a summary elements, which could include things like a total for
> the elements of that group as well as a running total for all the
> groups seen so far.  You might also want want to break the list of
> elements into chunks of 50 or a pageful, and add a running total at
> the end of the page?
> If so, there is the "Grouping" example in the XQuery specification
> and probably some in the use cases that might be of interest. 
>  However,
> they don't seem to be quite appropriate - at the very least they would
> be difficult to implement efficiently.  (I assume you want to be able
> to make a single pass over the data, without storing it all in
> memory as a DOM.)
> 
> As an implementor (www.qexo.org) I've done some very preliminary
> thinking and work on single-pass processing of a subset of xqueries.
> (The query 'document("mydoc.xml")/book/chapter)' is automatically
> processed by my implementation in a single pass without creating an
> in-memory DOM.)  It should be possible to generate more inteesting
> reports, including group summary elements, in a single pass, but I
> haven't really investigated it yet.
> -- 
> 	--Per Bothner
> per@bothner.com   http://www.bothner.com/per/
>

Received on Thursday, 24 January 2002 13:05:42 UTC