This is the mail archive of the
java@gcc.gnu.org
mailing list for the Java project.
Re: GCJ and DK compliance stats?
Tom Tromey <tromey@redhat.com> writes:
> >>>>> "Kevin" == Kevin A Burton <burton@openprivacy.org> writes:
>
> Kevin> Are there any tools for comparing GCJ to the exiting JVMs?
>
> There aren't any public ones that I'm aware of.
OK. Good. I will write one. Actually it will also help me get a grip on the
GCJ anyway :)
> Kevin> It seems to me that it would be very easy and would only take
> Kevin> me a few hours to write reports for this.
>
> One question is the legality of it. But if you restrict yourself to public
> and protected fields, methods, and classes in the java.* and javax.*
> namespaces you'll probably be ok.
I don't think there is a legal issue here? Maybe we are thinking about
different implementations though.
Basically your run an export on the current VM, reflect its core classes, dump
to a .dat file.
Then you can compare this to the core VM. The only thing we would publih would
be the stats which I think would be OK...
> Another important point is to make sure it ignores certain things. For
> instance it should ignore whether a given method is implemented in class X or
> in X's superclass
hm. good point.
> -- this difference is typically an unimportant implementation detail. Also it
> should ignore when unchecked exceptions are declared in a `throws' clause, as
> this is a meaningless difference.
Perhaps. What do you mean by uncheck exceptions? AKA RuntimeExceptions?
> If you write such a tool, and have it generate HTML,
initially it will be text...
> I'd like to put the results on the gcj web page.
that is what I was thinking. :)
> Eventually I'll have the Classpath comparison done automatically; we could do
> the same for the JDK comparison (but I think we would need a way to do this
> without actually installing the various JDKs on gcc.gnu.org).
We could do static exports of each proprietary JDK that we know about and then
run the reports from the .dat
> Kevin> This would give us some stats to post to the website so that others can
> Kevin> understand how complaint GCJ is WRT other VMs.
>
> With the exception of AWT, I believe we're actually pretty close to 1.2.
And swing... right?
BTW. I was thinking about calling it 'jvmdiff'...
Kevin
--
Kevin A. Burton ( burton@apache.org, burton@openprivacy.org, burtonator@acm.org )
Location - San Francisco, CA, Cell - 415.595.9965
Jabber - burtonator@jabber.org, Web - http://relativity.yi.org/
Did you hear why they're using Windows 2000 as a prison guard? Because it
always locks up!