Re: [xslt] Test Suite question

[Daniel Veillard <veillard redhat com>]

  Hi William,

On Fri, Mar 16, 2001 at 07:25:50PM +0800, William M. Brack wrote:
> We are currently working on numerous problems within transform.c, mostly concerning the "position" function (because proximityPosition and contextSize are frequently irrevocably modified by subroutines) and parameter passing. 

  I didn't investigated the parameter passing problem you raised yet.
Concerning the proximity position and context size, I agree there are some
grey areas where they may not be computed correctly. I'm very grateful you
can chase those ... Please indicate if your changes are likely to affect the
XPath part too
  
> We are progressing well, but in a previous exchange it was suggested that we should "make test" to confirm that changes we make do not adversely affect the system.  It seems to us that the "expected" results from the "make test" are, in fact, incorrect for many of the tests (especially those related to the "docbook" function).  In particular, we look at the stylesheet and corresponding xml file to see what we would expect to be produced, then we run the Saxon xslt implementation on the files and confirm that what we expect is what is produced.  Then we look at the "make test" 'expected' result (in tests/docbook/result/html) and see some very substantial differences.

> 
> So, my question: where do the 'expected' results in test/docbook/result/html come from, and how confident are you that they are, in fact, correct?

  I'm very confident that they indeed reflect problems !
But I think the other tests should be right (unless I misunderstood
something) because I built, ran and checked them each individually.

Daniel

-- 
Daniel Veillard      | Red Hat Network http://redhat.com/products/network/
veillard redhat com  | libxml Gnome XML XSLT toolkit  http://xmlsoft.org/
http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/