RE: Junit-type Framework for XSLT 2 Functions and Tem

FWIW, the XQuery WG has a large set of tests under development, where the
test cases are described by metadata in an XML catalog. My driver for these
tests is written in XSLT, essentially as a transformation from the test
catalog to the document that reports the test results. It does rely on
extensions: notably the ability to run a query from a stylesheet, and
importantly the ability to catch errors.

Michael Kay
http://www.saxonica.com/ 

> -----Original Message-----
> From: Eliot Kimber [mailto:ekimber@xxxxxxxxxxxxxxxxxxx] 
> Sent: 16 May 2005 15:42
> To: xsl-list@xxxxxxxxxxxxxxxxxxxxxx
> Subject:  Junit-type Framework for XSLT 2 Functions and 
> Templates?
> 
> I've started doing work with XSLT 2 and I'm finding that, with the new
> built-in function mechanism in XSLT 2, along with all the new built-in
> functions string processing and what not, that I'm writing more
> XSLT-level functions that I have in the past (in the past I would tend
> to write these functions as Java extension functions).
> 
> As these functions tend to be easily testable using automated tests I
> naturally created another test XSLT that provides unit tests for these
> functions. But it's not quite as convenient as having a real unit test
> framework ala JUnit.
> 
> I'm wondering if anyone has thought about the best way to implement a
> unit testing framework for XSLT 2? The approach I'm thinking about at
> the moment would use one transform to process a set of unit test
> templates in order to generate a new transform that would run 
> the tests
> and apply the results.
> 
> Another approach that might be productive would be to write a Java
> application that evaluates an input transform consisting of unit test
> templates and then applying those templates individually, under Java
> control (and in this case you could even make each template 
> call part of
> a real JUnit unit test, letting you take advantage of support 
> for JUnit
> in tools like Eclipse). But this would be a Java-centric 
> approach and it
> would probably be better to have something that is pure XSLT.
> 
> I haven't pushed on it, but it would be ideal if there was some way to
> apply the sort of introspection approach that JUnit 
> uses--that is, have
> a transform that processes itself in order to find the test case
> templates and run them, reporting success or failure. But I 
> don't think
> there's any way to do this directly (because there's no evaluate()
> function in standard XSLT.
> 
> Any ideas about how best to proceed?
> 
> Cheers,
> 
> Eliot
> -- 
> W. Eliot Kimber
> Professional Services
> Innodata Isogen
> 9390 Research Blvd, #410
> Austin, TX 78759
> (512) 372-8155
> 
> ekimber@xxxxxxxxxxxxxxxxxxx
> www.innodata-isogen.com