Saxon

Yes, it's entirely possible that there have been minor changes in this area. We would need to see exactly what you are doing to be more precise.

Ok, I'm gonna try to extract all code...

java -cp "saxon-ee-10.9.jar" net.sf.saxon.Transform -T:com.jenitennison.xslt.tests.XSLTCoverageTraceListener -o:... -xsl:... -it:"{http://www.jenitennison.com/xslt/xspec}main" and many other system properties used by XSLTCoverageTraceListener

Java code is available here : https://github.com/xspec/xspec/blob/master/java/com/jenitennison/xslt/tests/XSLTCoverageTraceListener.java You should know this code, your are listed in contributors list

I can provide my whole code if required.

Best regards, Christophe

Sorry to disappoint you, but I don't know the code! I expect they acknowledged me because they started by hacking an example I had written to do something quite different.

I'll take a look though.

Thanks a lot, Michael ! Jeni or Florent did some hacking ? Strange ;-)

It seems the XSLTCoverageTraceListener is call, at least the open method at line # 72 because the root element is written.

Best regards, Christophe

Changes were made to tidy up the tracing information in Saxon 10: see https://www.saxonica.com/documentation12/index.html#!changes/spi/9.9-10 and search for "Tracing".

I'm a little surprised that it runs at all, rather than throwing some kind of Java error; the trace listener code supplied won't compile against Saxon 10.

There is no longer a trace call at the level that used to be LocationKind.XPATH_IN_XSLT. I imagine that change was made to avoid multiple calls at the same point, but I can see that it might be useful for a TraceListener to know that we're at the XSLT-XPath boundary.

Attached is a rewrite of the trace listener that compiles successfully; I don't know if it works as a plug-in replacement because I'm not sure how the data that's collected is subsequently processed.

Thanks a lot, Mike, for this !

It seems to work. I'm just surprised by the coverage result, xsl:sequence are all non-evaluated.

I'm gonna spend more time on this later, I teach today !

Attached a coverage report. You'll see very strange coloring...

Thanks again, Christophe

Is there any function inlining going on? It's worth checking what happens if you suppress all optimisations using -opt:0.