> Jon Eaves <[EMAIL PROTECTED]> wrote:
>
> >> > I read lots of emails in ant-user and ant-dev about bootclasspath
> >> > not working quite right.
> >>
> >> I think it is not always the same, but a bunch of different,
> >> possibly related problems.
> >
> > Well, that's why I asked the question .....
>
> OK 8-)
>
> >> > Was it broken ?
> >>
> >> It depends ...
> >
> > No, not really. Broken is either yes, or no. Like pregnant.
> > Neither can be "a little bit", or "it depends"
>
> It depends on what you expect it to do - does this sound better? If
> it doesn't do what you want, you may call it broken, but to others it
> is not.
Thanks for the clarification Stefan. I'll give you my definition of
broken.
"If it doesn't do what a javac from the command line does"
>
> What the bootclasspath attribute (successfully) does, is passing the
> content to the -bootclasspath argument of javac.
Which doesn't work (in this instance) because there are libraries
polluting the bootclasspath that needs to be "pure".
>
> > If I use the <exec> task with "executable=javac" and the _same_
> > options that are given to the <javac> task, it all works correctly.
>
> Because Ant runs javac in the same JVM as it is running itself. The
> only real solution probably would be to implement a forked javac
> variant for situations where you need a different bootclasspath for
> javac and Ant - this is not available yet.
Thank you. Much appreciated. Now I know that it doesn't work,
and I know that there isn't a fix, I can go down the path of
either getting the additional tasks, using "exec" or fixing it.
Cheers,
-- jon
--
Jon Eaves <[EMAIL PROTECTED]>
http://www.eaves.org/jon/
