Re: Measuring the complexity of XSLT stylesheets

Jirka makes the following excellent point:

-----------------------------------------------------

I think that many people think about XML as about hierarchical way of 
describing/storing data. XSLT by default traverses XML data from the 
root to the leaves using templates. If you need to access other axes 
than child and descendant in your stylesheet, then there is usualy some 
complexity behind your data from human perspecitive of view. So counting 
number of XPath expressions that use different axes than child and 
descendant might give you some metric not completely unrelated to 
complexity of input data and their processing. Or you can assign 
different weight to differents axes to make it more "scientific".

-----------------------------------------------------

It seems to me that, when you have to pull data from different parts of 
the hierarchy or output something that isn't based on the source 
structure, you've got a big jump in complexity. These kinds of problems 
are usually addressed by Meunchian grouping. Thus, if the stylesheet 
contains a key, I would add a multiplier to its complexity. You might add 
another multiplier for variables that contain result tree fragments, 
though a lot of people do that unintentionally.

I strongly suspect that assigning a weight to the length of match 
expressions and then multiplying by some value for the presence of one or 
more keys would give you a good measure of a stylesheet's complexity, at 
least from the point of view of writing and maintaining the stylesheet.

Jay Bryant
Bryant Communication Services
(presently consulting at Synergistic Solution Technologies)