update on socket problem



I spent some time working on the "socket problem" today. I made some
progress, but I do not yet have the solution. 

The simplest way to invoke the bug is to type
	load("foobar");
or other filename that does not exist.

It seems that the maxima process is dying because the function merror
invokes
	(throw 'macsyma-quit t)
The throw is caught when maxima is invoked from the command line, but it
is not caught when maxima is invoked by xmaxima.

I still do not understand why.

--Jim