[GRASSLIST:2839] Re: d.mon can't bind to socket

[Eric G. Miller]

On Thu, 3 Jan 2002 15:20:51 +1100, Anthony Morton <abmorton at mac.com> wrote:

> 
> Hi GRASS people,
> 
> I think this problem has been reported before but I haven't seen a
> response.  (Maybe I'm not looking in the right place.)
> 
> I've just installed grass5.0pre2 under MacOS X using XFree86 4.1 with
> XDarwin 1.0.6.  All goes well until I try and start a monitor.  Then I
> get the following:
> 
> GRASS 5.0.0pre2 > d.mon start=x0
> using default visual which is TrueColor
> ncolors: 16777216
> Can't bind to socket; already in use?
> No socket to connect to for monitor <x0>.
> Problem selecting x0.  Will try once more
> No socket to connect to for monitor <x0>.
> 
> Thus I'm prevented from starting any monitors.  Has anyone any
> experience with this problem?

Do you get a directory /tmp/grass-$USER (where $USER is your login)?
And if so, does it contain anything? (like a file called "x0").

It would seem that this could only occur if,  the GRASS monitor is
not able to delete a previous /tmp/grass-$USER/<socket file> or if
UNIX/LOCAL sockets are not implemented.  Perhaps a permissions
problem on /tmp ?  Otherwise, I wouldn't expect to see the
"Can't bind to socket" message.  The system should have already
verified/created /tmp/grass-$USER/ or you should've received a
different error.  One other, unlikely, failure scenario is if
the filesystem "/tmp" is not a local "mount point" (in which
case, the implementation of "LOCAL" sockets is not required
to work -- though it may).  But, there isn't any good reason
to have /tmp mounted on a networked filesystem, so that seems
unlikely.

As others clearly have gotten the GRASS monitors working fine
under OS X,  it's hard to say what the problem is.  In the
past, errors about creating monitors have often been the
result of a build configuration mismatch (where part of GRASS
was built thinking it'd be using the older FIFO communication
method, and another part thinks it's using the now standard
UNIX Sockets communication method).  But, in this case, it
appears all ends agree... So, ???

-- 
Eric G. Miller <egm2 at jps.net>