Feedback request: OpenJDK Community Innovator's Challenge Grants

[Andrew Haley]

Andrew John Hughes writes:
 > On 07/01/2008, Andrew Haley <aph at redhat.com> wrote:
 > >
 > > Andrew John Hughes writes:
 > > > On 07/01/2008, Andrew Haley <aph at redhat.com> wrote:
 > > > >
 > > > > Arnd-Hendrik Mathias writes:
 > > > >
 > > > > > How about some concept like using some java2native compilers or
 > > > > > some java enabled gcc to build the necessary set of bootstrap JDK
 > > > > > components in a pre-build-step and then using these for building
 > > > > > the real OpenJDK.
 > > > >
 > > > > That's how IcedTea works.
 > > >
 > > > It doesn't do anything native, AFAIK -- it just creates a
 > > pseudo-bootstrap
 > > > JDK.
 > >
 > > What does "it doesn't do anything native, AFAIK" mean?  I can't tell
 > > if you agree with me or not.
 > 
 > I interpreted the question (perhaps wrongly) as looking for a
 > solution that wouldn't require a Java VM and class library
 > i.e. that there would be a preliminary step that created a native
 > toolchain of javac, etc. which could then be used to bootstrap the
 > OpenJDK in full. IcedTea only does half of this; it creates the
 > pseudo-bootstrap environment using ecj and a Java VM, but this is
 > not native.

I still don't know what you mean by this.  The OP said "use some java
enabled gcc to build the necessary set of bootstrap JDK components in
a pre-build-step."  That's exactly what we do with gcj!

 > I'm not sure how much of an issue that is, as you'd still need
 > libgcj to run the native binaries.

None at all, I wouldn't have thought.

Andrew.

-- 
Red Hat UK Ltd, Amberley Place, 107-111 Peascod Street, Windsor, Berkshire, SL4 1TE, UK
Registered in England and Wales No. 3798903