Re: Failure in building Emacs 23.0.50 on Cygwin (fwd)

[Dan Nicolaescu]

Angelo Graziosi <address@hidden> writes:

  > On Tue, 4 Sep 2007, Dan Nicolaescu wrote:
  > 
  > > "d.henman" <address@hidden> writes:
  > >
  > >   > Angelo,
  > >   >
  > >   > content wise it looks good to me, but I think that changing it to the 
following
  > >   > format, same content as your patch (see modifed patch below ), but 
this format makes it easier to understand and would give it more robustness for 
future mainenance and any possible changes and extensions for CYGWIN specific 
issues.
  > >   >
  > >   > Also I would like to wait and get a response from the cygwin 
developers, before any commit is made.
  > >   >
  > >   > Regards,
  > >   >   Darel Henman
  > >   >
  > >   > ----------------------------------------------------------------
  > >   > --- term.c    2007-08-30 06:46:50.000000000 +0900
  > >   > +++ /tmp/term.c.fixed 2007-09-03 09:15:22.915209300 +0900
  > >   > @@ -3131,6 +3131,10 @@
  > >   >        setpgrp ();
  > >   >        no_controlling_tty = 1;
  > >   >  #else
  > >   > +#ifdef CYGWIN
  > >   > +      setsid ();
  > >   > +      no_controlling_tty = 1;
  > >   > +#else
  > >   >  #ifdef TIOCNOTTY                /* Try BSD ioctls. */
  > >   >        sigblock (sigmask (SIGTTOU));
  > >   >        fd = emacs_open ("/dev/tty", O_RDWR, 0);
  > >   > @@ -3145,6 +3149,7 @@
  > >   >        /* Unknown system. */
  > >   >        croak ();
  > >   >  #endif  /* ! TIOCNOTTY */
  > >   > +#endif  /* ! CYGWIN */
  > >   >  #endif  /* ! USG */
  > >   >      }
  > >   >  #endif
  > >
  > >
  > > Just my curiosity, does this patch make emacs work on cygwin?
  > > If it doesn't, is this function called before things go bad?
  > 
  > 
  > IF 'setpgrp' is similar to 'setsid'
  > (http://www.opengroup.org/onlinepubs/009695399/functions/setsid.html) AND
  > IF Cygwin list suggests to use 'setsid' to dissociate TTY
  > (http://cygwin.com/ml/cygwin/2007-09/msg00026.html) THEN
  > 
  > the above patch is quite natural... OR NOT?

You have not answered the fundamental question: does the patch work? 
I don't think this code path is exercised if you run in a window
system, it will probably only be used if you use "emacsclient -t" or
"emacs -nw".