Re: JOS

[Brian Jones]

address@hidden writes:

> Brian Jones (address@hidden) wrote:
> > The following article was interesting to me.  I don't know what the
> > state of the JOS project is or who our liason is.  Could someone from
> > that project bring us up to date with respect to this email referenced
> > below?
> > 
> > http://www.jos.org/pipermail/arch/1999-December/000358.html
> 
> It is interesting.  If I might extract their points:
> 
> 1.  They think that replacement classes should be written in an alternate
> namespace, then ported over when completed.  This certainly solves some
> chicken and egg problems, but it creates new ones of its own.  I don't
> object to that approach, but what we currently have seems to be working
> fine.

Point (1) was interesting in that it might be easier to test code
against a more reliable VM, making it much easier to just assume the
problem is in gcp (GNU Classpath) until proven otherwise.  I'm
somewhat eager to take the VM out of the picture when trying to debug
the 98% (swag) of gcp which doesn't have a VM specific hook.

Unlike the article however, I'd prefer to keep CVS code in the current
package name space and look at using a java parser to change the
package names of user defined parts of the tree for testing.  Has
something like this already been written elsewhere?

Brian
-- 
Brian Jones <address@hidden>