Another Make
Cedric Berger
cedric@wireless-networks.com
Tue Oct 10 16:05:00 GMT 2000
>> On 2000年10月10日, Cedric Berger wrote:
>> > The same as you: redoing Ant, which is moreover provided with such
> > a liberal license, is inherently stupid!
> > We, developers, don't want/like/need fragmentation.
> >
> > Beside: If you really want a GPL Ant, there is an easier way to redo
> > anything:
> > - Download Ant source
> > - Rename it Dog
> > - Stick the GPL license on it
> > - Give a few due acknowlegments
> > - That's it.
> >
> > This is the beauty of the BSD license.
> >
> > Of course, if you feel like reprogramming Ant is a good and fun
> > exercise for you, do it; but please don't publish the result,
> > unless it is really really better.
> >
> > Sorry for the flame,
>> No, no that's ok.
>> The only problem I have with Ant is that it depends on the latest
> Sun/parser xml stuff and on Apache/xerces xml parser.
>> Also this is deffinitively the right way of doing it, it might make it
> hard to port to gcj in the short term.
Trying to port it to GCJ would probably be a good thing to try, and
should (??) not be so difficult since Apache code is still JDK 1.1-based
if I'm not mistaken. Where it fails, you can either:
- help fix GCJ (which is the ultimate, long-term goal)
- compile from bytecode (to avoid inner-class front-end issues)
- tweak Ant
If the Xerxes/parser is too big, you can certainly write a scaled-down
version with just the features needed by Ant for now, without much error
checking, keeping the same interface.
> I was thinking of a simpler, cut down, version which would use the same
> build.xml format that Ant. So that simple build.xml scripts could be use
> as part of gcj for now and probably move to the more powerful Ant when
> it does port to gcj.
If all the above doesn't work, I guess it may then makes sense.
Cedric
More information about the Java
mailing list