RFR: 8054717: SJavac should track changes in the public apis of classpath classes!

Erik Joelsson erik.joelsson at oracle.com
Tue Mar 3 11:53:33 UTC 2015

That combination didn't actually build for me. When compiling 
jdk.jconsole, the following happened:

com.sun.tools.javac.code.Symbol$CompletionFailure: class file for 
java.awt.datatransfer.UnsupportedFlavorException not found
     at com.sun.tools.javac.api.JavacTaskImpl.doCall(JavacTaskImpl.java:92)
     at com.sun.tools.sjavac.comp.SjavacImpl.compile(SjavacImpl.java:129)
     at java.util.concurrent.FutureTask.run(FutureTask.java:266)
     at java.lang.Thread.run(Thread.java:744)
Caused by: com.sun.tools.javac.code.Symbol$CompletionFailure: class file 
for java.awt.datatransfer.UnsupportedFlavorException not found

 From what I can tell, jdk.jconsole uses java.desktop, which uses 
java.datatransfer (and re-exports it), where the missing class is. My 
best guess is that something in sjavac (API checking?) needs to traverse 
down into the transitive dependencies. This might be wrong, but as a 
workaround, it's easiest to fix in the makefiles so that we can have a 
working sjavac enabled jdk build.

Here is an updated webrev which adds 3 levels (should cover everything I 
think) of transitive dependencies on the classpath when compiling modules:



On 2015-03-03 12:17, Erik Joelsson wrote:
> Hello,
> Here is my suggestion for makefile changes to go with the sjavac 
> changes. Adding build-dev to get review for my part.
> Webrev: http://cr.openjdk.java.net/~erikj/8054717/webrev.root.01/
> Bug: https://bugs.openjdk.java.net/browse/JDK-8054717
> /Erik
> On 2015-03-02 13:31, Andreas Lundblad wrote:
>> On Mon, Mar 02, 2015 at 09:42:58AM +0100, Erik Joelsson wrote:
>>> Sounds good to me. So you think I should do the makefile changes to
>>> adapt the JDK build to this patch in a separate bug?
>> I'm not sure what's best, but I don't think we need a separate 
>> bug-entry.
>> If you give me a pointer (or patch) on how to create/clean up a dummy 
>> empty bootclasspath directory, I can do the push.
>> -- Andreas

More information about the compiler-dev mailing list