Re: JNI_MD.H

[Christopher Abbey <cabbey at bresnanlink dot net>]

Today, Veit, Michael wrote:
> I am trying to compile a makefile using the latest version of Cygwin. The
> file that is getting compiled is a JNI implementation file. When I compile I
> get "deferencing pointer to incomplete type" errors all over. Don't know if
> this a common C error, or a cygwin problem. I'm a newbie at this. forgive my
> ignorance.

I've tried to do this with several java sdks, as yet I've had no success.

Most implementations' headers seem to assume MSVC (even IBM's) and will
not compile correctly with gcc. There are patches from Mumit Khan at
http://www.xraylith.wisc.edu/~khan/software/gnu-win32/README.jni.txt
that will get you part way (you'll need to repeat the same basic step
a few more times than the once or twice his docs indicate). However,
as far as I can tell this will only lead you to frustration as the
resultant DLL will cause most JVMs to segfault immediately on load.

Note that I've long since given up on this, and things may have
improved some by now, but I doubt it. Either way, the URL above on
xraylith is the best resource I've seen on this topic. What we really
need is a JVM vendor to adopt cygwin as a platform and "port" their
JVM and sdk to it. I happen to be in a position to ask one vendor to
consider this... what would people think of that? would it be welcomed
by "the cygwin community"? (realize that this would be a closed source
binary only - there is no way around that. :(  ) or would it be greated
by the usual derision, ridicule, scorn, and ceremonial rock throwing?

-- 
now the forces of openness have a powerful and
  unexpected new ally - http://ibm.com/linux


--
Want to unsubscribe from this list?
Check out: http://cygwin.com/ml/#unsubscribe-simple