[cmake-developers] [review] add_jar (UseJava) uses cmake_parse_arguments

[Andreas Schneider]

On Tuesday 26 March 2013 10:33:03 Andreas Schneider wrote:
> On Monday 25 March 2013 13:14:33 Brad King wrote:
> > On 03/25/2013 12:28 PM, Matthew Woehlke wrote:
> > > these must now be explicitly specified with INCLUDE_JARS. Other named
> > > arguments are SOURCES (optional, to 'reset' to listing sources after
> > > 'INCLUDE_JARS'), OUTPUT_DIR, OUTPUT_NAME, VERSION and ENTRY_POINT. These
> > > supersede the old variables, which are still supported for backwards
> > > compatibility.
> > 
> > That looks like a nicer interface to me.  Andreas, Nicholas, does it
> > work for your use cases (and not break existing projects)?
> 
> Yes, that's true.
> 
> > > I'm on the fence if this should target 2.8.11. On the plus side, it
> > > means the historic behavior of ignoring jar files listed as sources will
> > > be preserved. On the down side, it is late in the cycle and this is a
> > > less trivial change.
> > > 
> > > If it does not go in, I will redo the change to preserve the behavior
> > > from the previous commits (i.e. jars and targets are accepted in the
> > > sources list).
> > 
> > This is one reason I don't like rushing such changes in during the
> > RC cycle.  Now we have to decide whether to rush an even bigger
> > change in, revert the original, or have behavior that was preferred
> > for exactly one release and have to support it forever.
> 
> I would prefer this change before we define/add more types of the function
> and need to support more backward compatibility.
> 
> It is a pain now. But better than more pain later :)
> 

Please take a look at

http://cmake.org/gitweb?p=stage/cmake.git;a=shortlog;h=refs/heads/fix_java_test

it fixes JNI detection on Linux Distributions, at least openSUSE and Fedora.


	-- andreas

-- 
Andreas Schneider                   GPG-ID: F33E3FC6
www.cryptomilk.org                asn at cryptomilk.org