Can't compile jdk15 on amd64 5.3-STABLE

[Conrad J. Sabatier]

On Sun, 13 Feb 2005 19:35:56 -0500, Nicolas Blais <nb_root at videotron.ca>
wrote:

> On February 13, 2005 19:30, David O'Brien wrote:
> > On Sun, Feb 13, 2005 at 02:33:42PM -0500, Nicolas Blais wrote:
> > > So far I've been running linux-sun-jdk-1.4.2.06_1 successfully on
> > > my box and works great in konqueror but not at all in firefox. So
> > > I tried to compile jdk15 and it gives me the following error:
> >
> > ...
> >
> > > I googled around and found
> > > http://lists.freebsd.org/pipermail/freebsd-java/2005-February/003446.html
> > > and several replies but no real fix for me. Apparently I'm not the
> > > only one in this situation.
> > >
> > > From what I understand, the command "javac" that is executed
> > > before the fail point is 25814 characters long and on my
> > > 5.3-STABLE machine I max out at 16384 characters (according to
> > > many configure scripts).
> >
> > I'm willing to MFC what ever is needed to make this work, but I
> > really can't parse this.
> >
> > You can build jdk15 on FreeBSD/AMD64 6-CURRENT, right?
> > You cannot build jdk15 on FreeBSD/AMD64 6-CURRENT, right?
> > You feel this is due to a difference in the maxium command line size
> > (in characters), right?
> 
> Actually, I can build jdk15 on a amd64 6-CURRENT machine and I can't
> on a  amd64 5.3-STABLE. Yes I believe  (as well as others) that it is
> due to a  maximum command line size (as mentioned, the javac command
> during that point  in the makefile is about 25000 char long and on my
> -STABLE I can only get  16384)

The same problem occurs in CURRENT for amd64.  Several people in
the java list pointed out a workaround just to get the thing built,
but this is not a proper solution.

I offered Alexey use of my Athlon 64 box, but so far, he hasn't logged
in here.  Apparently, the problem doesn't affect Opteron boxes for some
strange reason, only Athlon 64, so he wasn't able to reproduce the bogus
behavior himself.

Even after employing the workaround to get the build to complete, and
installing the port, the bad news is that the resulting JVM is all but
useless, gobbling up absolutely insane amounts of RAM (no less than one
gigabyte to run even the most trivial app).

Very disappointing, and very frustrating as well, to say the least.

-- 
Conrad J. Sabatier <conrads at cox.net> -- "In Unix veritas"