harmony-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Mark Hindess <mark.hind...@googlemail.com>
Subject Re: harmony & freebsd
Date Sat, 27 Feb 2010 09:36:12 GMT

In message <201002240929.o1O9TJXi028098@d06av04.portsmouth.uk.ibm.com>,
Mark Hindess writes:
> In message <4B84A88E.3070301@p6m7g8.com>, "Philip M. Gollucci" writes:
> >
> > Hi harmony!,
> Hi!
> > We might be looking to start using harmony on people.apache.org
> > (minotaur) after we update to stable/8 since the freebsd port
> > java/jdk16 is having compile issues right now.
> >
> > Compile log is available here:
> > http://tb.apache.org/errors/8-STABLE-people/jdk-
> >
> > If you have any insights into the above issue please reply to me
> > directly or on infrastructure@.
> We can't really help with that.  Since Harmony contributors are by
> definition not permitted to look at details of the non-ALv2 compatible
> implementations of java and looking at compile logs might expose us to
> such details.
> > Additionally, we'd be interested in knowing how well supported harmony
> > is on FreeBSD 8 and how well it would work with ant, anika, and
> > velocity [aka be able to build www.apache.org and others at minimum]
> I have no idea about the current status.  Way back in 6.3 days I did
> get drlvm/classlib to build on FreeBSD but there were thread library
> issues (with either of the three thread library choices at the time)
> that prevented it from running.  I don't recall clearly but I think
> this had to do with recursive mutexes.  The details will be in the mail
> archives.
> If I get a moment, I'll set up a FreeBSD 8 system and take a look at
> the current state of things.

I updated the freebsd icu dependency downloads so now harmony builds on
FreeBSD 8.0 x86 and x86_64 straight out of svn.  It seems to run with
-Xint but not with the jit.  There are some fundamental issues with
sockets but otherwise the class library behaviour looks pretty good.

I've fixed one socket issue (for some insane reason "(new
Socket()).connect(addr)" does a bind which current platforms don't seem
to mind but it is wrong and FreeBSD rightly complains).  I'll check this
fix in after the code freeze but if anyone wants the fix let me know and
I'll attach it to a JIRA.

This clears up client socket issues but there are some server socket
issues with bind syscalls being called with the wrong address.  I'm
looking at these now.

Once I've cleared up the classlib issues I'll try to find time to
investigate why running with the jit doesn't work.


View raw message