Re: problems building on OSF1

[Anselm Helbig]

At Sun, 29 Oct 2006 11:03:47 -0500,
Eli Zaretskii <address@hidden> wrote:
> 
> > Date: Sat, 28 Oct 2006 23:19:02 +0200
> > From: Anselm Helbig <address@hidden>
> > 
> > I tried to build the emacs-22.0.90 pretest on OSF1 today. 
> > 
> > uname -a
> > => OSF1 helene.molgen.mpg.de P4.0 227 alpha
> 
> Thanks for testing.
> 
> > The problem seems to be that the version of OSF1 I am using does not 
> > support IPv6. Yet to me it looks like emacs assumes that all systems 
> > other than Windows NT support IPv6. 
> > 
> > 124 /* Disable IPv6 support for w32 until someone figures out how to do it
> > 125    properly.  */
> > 126 #ifdef WINDOWSNT
> > 127 # ifdef AF_INET6
> > 128 #  undef AF_INET6
> > 129 # endif
> > 130 #endif
> 
> No, Emacs doesn't assume that.  Rather, the configure script probes
> for the IPv6 support and defines AF_INET6 accordingly.  Could you
> please look into config.log and see where that test misbehaves, and why?

I don't see IPv6 mentioned during the configure process. The checks
that are closest are these:  

  checking for socket... yes
  checking netinet/in.h usability... yes
  checking netinet/in.h presence... yes
  checking for netinet/in.h... yes
  checking arpa/inet.h usability... yes
  checking arpa/inet.h presence... yes
  checking for arpa/inet.h... yes

and that's all. I grepped the source tree for AF_INET6 and found it
only in src/process.c and src/Changelog. Frankly, I don't know what
else to look for. I hope I did mention that the code snippet I pasted
before was from src/process.c? 

> > Loading subr...
> > Symbol's value as variable is void: nil
> > 
> > I have no clue how to debug this.
> 
> One way of debugging this is run temacs under a debugger, put a
> breakpoint at the functions Ferror and Fsignal, then let temacs run,
> and when the breakpoint breaks, see where exactly this error happens.
> If you use GDB, then the command xbacktrace should display a
> Lisp-level backtrace that should pinpoint the offending place.  (You
> need to invoke GDB from the src directory, where the file .gdbinit
> defines the xbacktrace command.)  It is best to rename subr.elc so
> that Emacs will load subr.el, because the backtrace info will be
> more readable then.

Thanks for your instructions. I tried to follow them but was not able
to set a breakpoint at "Ferror" (Fsignal worked fine, gdb version is
5.2.1). I am not able to find Ferror in the sources as well. And
breaking at Fsignal does not seem to be sufficient, temacs just
terminates with an error without reaching the breakpoint at Fsignal. 

I'm using the tarball from 

  ftp://alpha.gnu.org/gnu/emacs/pretest/emacs-22.0.90.tar.gz

which I downloaded yesterday, 2006-10-28 at 21:00. 

I would be glad to provide more information, but I need some
guidance. My knowledge of C is quite basic and therefore didn't study
the C-side of emacs yet.

Regards, 

Anselm