Stefan Bodewig wrote:
On Thu, 05 Feb 2004, Steve Loughran <[EMAIL PROTECTED]> wrote:


The one interesting issue is that as Java1.5 beta has just shipped,
we should test against that to see if there are any surprises that
we can fix in 1.6.1.


;-)

<http://stefanbodewig.blogger.de/stories/50816/>
<http://marc.theaimsgroup.com/?l=ant-dev&m=107528601129819&w=2>
<http://stefanbodewig.blogger.de/stories/59918/>

Ant 1.6.0 doesn't allow you to use source="1.5" in <javac>, 1.6.1
will.

At least on the Linux version, the -target switch for javac is broken
(in the JDK, not in Ant) as -target 1.1 will produce 1.4 classes
anyway.  And the default value for -source seems to have changed as
the JDK complains about -target 1.1 unless you also explicitly set
-source 1.3.  It looks as if 1.4 was the new default.

I just don't want to adapt anything here before I see the docs for JDK
1.5 as it is hard to tell what has been changed intentionally and what
is a bug in the JDK.

I have successfully bootstrapped Ant with 1.5 and ran the testsuite.
I didn't try the javadoc task (but that should be trivial) or rmi
(hmm, could try to build JMeter).  I don't have any project that uses
javah, though.


we have some sample code from the ant book; I could try that out. Actually, we could make that part of the test suite -we dont need to verify that javah generates compilable C++, only that it generates C++.


Another area to think about is inheritance of stuff on the classloaders. Java1.5 does a good thing in that xerces and xalan have been repackaged under Sun, so you can have a real xerces and xalan without suffering endorsed directory grief (finally!). This may change our actions regarding which bits get passed down from above, and which bits dont


--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]



Reply via email to