Would it surprise you to learn that
returns the empty sequence when -T is used and returns the correct node
when it is not?
Be seeing you,
Norman Walsh <ndw@...> | The Future is something which everyone
http://nwalsh.com/ | reaches at the rate of sixty minutes an
| hour, whatever he does, whoever he
| is.--C. S. Lewis
From: Michael Kay <mike@sa...> - 2005-04-05 18:27:19
> Would it surprise you to learn that
> <xsl:variable name="spanspec"
> returns the empty sequence when -T is used and returns the
> correct node when it is not?
Well, it shouldn't happen, but it doesn't surprise me that greatly. Tracing
generates extra nodes in the compiled expression tree, and therefore affects
subsequent rewrites of the tree by the optimizer; and I run thousands of
tests with tracing off and very few with it on, so such problems are,
unfortunately, quite possible.
Please send a repro if you can.
From: Michael Kay <mike@sa...> - 2005-04-06 14:09:44
Thanks, problem identified. When tracing is switched on, the current()
function isn't being evaluated at the outermost level of the XPath
expression. Applies to both 8.3 and 8.4.
> -----Original Message-----
> From: saxon-help-admin@...
> [mailto:saxon-help-admin@...] On Behalf Of
> Norman Walsh
> Sent: 05 April 2005 19:49
> To: saxon-help@...
> Subject: Re: [saxon] -T behavior?
> | Please send a repro if you can.
> Document and stylesheet attached. Using 8.4 with -T it produces
> the xsl:message
> count of nodes 0
> without -T it produces
> count of nodes 1
Get latest updates about Open Source Projects, Conferences and News.