[JDEV] Re: Trouble compiling Jabberd v1.4.2 under recent versions of Cygwin

Frank Seesink frank at mail.wvnet.edu
Tue Jun 17 00:01:48 CDT 2003


	Apologies.  It's late, I'm tired, and I just went over my post and 
realized some typos.

Frank Seesink wrote:
...
> And copying javadns.exe to /usr/local/bin just made it more accessible
> to jabberd.exe when you fired it up.

	CORRECTION:  jabadns.exe (not javadns.exe)

...
> * I have tried the above using both the original Jabberd 1.4.2
>   distribution files (making the one change to ./jabberd/jabberd.h
>   mentioned earlier) and using my modified Makefile setup.  In all
>   cases the DLLs and .EXEs built fine.  And getting jabberd.exe to
>   run to give me version information works fine.  But the moment I try
>   to bring the server online, that's it.

	CORRECTION:  The one file I changed from the original distro was 
./jabberd/Makefile (not jabberd.h).  And I tried doing just that, OR I 
did the default distro install, then threw down my modified ./configure 
and 'Makefile's, and did the install with that.  No matter which method 
I used, no matter which version of GNU pth was installed, the end result 
was always this STACK OVERFLOW/segmentation fault error.

	I know the code is compiling, but I can't help but wonder what is 
causing this.  gcc?  ld?  cygwin1.dll?  So many places where things can 
change.  I just spent the better part of a month tracking down various 
bugs in PostgreSQL under Cygwin.  I would very much like to do the same 
here, both for myself, and hopefully to help get Jabberd to the point 
where maybe there can be a Cygwin package as part of Cygwin itself. 
Anything to help bring more people over to JABBER.  Those that can 
afford TIMP, Merlin, Antepo, etc., hopefully will support those 
endeavors.  But in our particular case, I have to first prove JABBER's 
usefulness before there's any chance of getting funding. :-/

	Anyway, any insight into this error message would be greatly 
appreciated.  I'm about out of ideas at this point.

P.S.	Yes, all the pth tests from 'make test' complete successfully...
	yet one more thing that makes this so annoying.





More information about the JDev mailing list