Here are some draft thoughts on Ant1.7.1.
First, what is the goal? Bugs only, or bugs+features?
Refresh dependent libraries (xerces &c)
Planning: ApacheCon + email lists: may 2007
Java 1.6 support
XavierHanin: I do have access to Vista Professional edition
SteveLoughran: I have a VMWare image of Vista Enterprise Edition.
JanMatèrne: I'll set up a VMWare image of Vista Ultimate Edition (32bit, German)
[ http://issues.apache.org/bugzilla/show_bug.cgi?id=41958 Javadocs breaking]
* Better proxy support on Java 5+
XavierHanin: Difficult to say, it depends on external factors. The major point for the moment is to recruit new committers.
make <scriptcondition>, <scriptselector>, etc, more useful through : attributes and elements, re-use
(done for <scriptdef> as of 1-may-2007)
BigProject logger with more logging of which project we are in (entry, exit, full location on an error), quiet on targets with no output, and in -verbose mode, log which build file a target actually came from. Maybe print finish time too.
add an UnexpandedString() datatype that can be used in attribute assignments for classes that don't want properties expanded...we could use this in macro processing to prevent double expansion of properties. (this would fix a few bug reports too)
produce an official izpack containing Ant+core+apache libraries (Xalan, BSH, &C). And JUnit?
= Postponed For Ant 1.7.2 ==
add an <xmlmessage> element for mail that takes an XML message, so you dont need to CDATA escape XHTML messages.
add a <antfork> task that runs a completely new ant process, with different env, maybe even JDK. Sometimes people need this; running forrest builds is just one example. We'd use <java> to set it up.
add support to <subant> a graph of targets, not just a list, so it can skip all dependents of a failing build
SCM: Microsoft Visual SourceSafe Tasks + SourceOffSite
SCM: Cvs + CvsChangeLog + CvsVersion + CVSPass + CvsTagDiff
implement something similar to <functionaltest> that steve uses in smartfrog for starting smartfrog and running tests against it.
allow <waitfor> to fail if it times out (<functionaltest> will need this)
this may imply we need a way to get one resource to resolve a relative link to another resource...a new interface.