"Aaron M. Renn" wrote:
> Paul Fisher ([EMAIL PROTECTED]) wrote in message 72:
> > I'm being forced to come out of my AWT/JDK development hole to do work
> > on Japhar w/ Classpath.
> >
> > Now, if I could only get things to run.... :)
> >
> > Japhar is segfaulting on me -- in pop_frame.
> >
> > So, before I spend a few hours tracking down the problem, is
> > Japhar-CVS (current as of now) known to work with our 0.0 release?
>
> The current Japhar-CVS does not work with Classpath 0.00. I just
> tested it. It bombs for me with:
>
> Fatal Java VM Error: ThrowNew unable to allocate message
> Bus error
>
> This is running my BufferedInputStreamTest.java file, which I can
> send anyone who is interested. Japhar+JDK1.1 works fine on it.
Is this in Mauve? if so, I'll take a look. otherwise, can you send it
along to me?
> I tracked it down a little bit. ThrowNew calls NewStringUTF, which
> tries to allocate a character array. A call to NewCharArray for a
> length of 10 (for the string "returnType") is returning null for
> some reasons. Any clue guys?
Not right off the bat. i'll take a look now.
> I must confess, I was still running Japhar 0.07. I'll try to stay
> more current from now on. When we fix this problem is it possible for
> the Japhar guys to put in a small regression test pass for a
> japhar+classpath combo to make sure things work after major changes?
yes. I think this is completely reasonable, and I'm sorry I haven't been
keeping up with classpath. I shall make more of an effort to do so -- i
mean, i run both jdk1.1 and 1.2 on my box. it wouldn't cause me any pain
at all to use classpath as well.
> BTW: I had two compile problems with the latest Japhar. First, it
> is looking for a file called nativeglue.h, which doesn't exist and
> doesn't get generated for me. I just touched it and all was well.
> Also, japhard is bombing bigger than life on readline, but I didn't
> try to get that working.
hmm... you on linux? could you send me the compile log?
xtoph
