Would anyone object to moving the passing StylesheetTestlets in
tests/bugzilla to the approprate functional test buckets, or to an
appropriate subdirectory of contrib in lieu thereof? Ideally, Bugzilla
and Jira should only contain issues currently being worked, and tests
which we pass should be part of the smoketest to guard against regression.
(Ought to confirm that these pass on Windows too before doing so.)
At least one of the failures I am seeing on Linux is a Windows/Linux
newline issue. That's an error in either how the processor is configured
(turn off prettyprint or explicitly configure newline -- I think the
latter is a supported option), or in how the results are compared
against gold, depending on how important the newlines are to what's
being tested. We should reconcile that. Lemme find that test again and
I'll open an issue.
A bunch of the bugzilla tests report as AMBG, since they don't come
with .out files (corresponding to gold files elsewhere). That may again
call for using JavaTestlets to validate the results.
I still need to look at whether one can just run both
StylesheetTestletDriver and JavaTestletDriver against a directory to get
combined handling of both categories which conforms to the existing
framework design. I'd like to avoid having both tests/contrib and
tests/contrib-java (for example) when the category calls for tests of
both sorts.
Just as a heads up: Many moons ago I wrote a set of articles on using
stylesheets to semi-automagically annotate stylesheets with logging
messages. There's one detail missing which could be cleaned up (I
cheated a bit on namespaces, since it hadn't occurred to me that the
right solution was to use a predicate to test the namespace URI in the
generated XPaths), but this may be useful in analyzing some failures,
assuming that we never got around to making a Xalan debugger hook.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@xalan.apache.org
For additional commands, e-mail: dev-h...@xalan.apache.org
