In message
<CAD_jPZ=ao_GHnVZMYSKgy6APzg=TL=jmsdpxxvvdcabwfde...@mail.gmail.com>,
Stephen Dennison <[email protected]> wrote:
>There's probably nothing wrong with localhost per se...
OK. I guess that's the good news.
>> Basically, once I start up X and fvwm, I see that all of my xterm
>> windows are automagically inheriting the following environment variable
>> and the value shown:
>How did you start up "X"? Which "X" are you running? What OS and are
>you using a login manager, display manager, etc?
I am using:
FreeBSD 11.0-RELEASE
xorg-7.7_3 (installed from packages)
I start X by just logging in (console) and then running xinit. I have a
(slightly) personally tailored .xinitrc file.
I am *not* using any X login manager.
>Perhaps you inadvertently set the value of DISPLAY in one of your dot
>files and that's the actual problem?
Ohhhhhhhh! Good lord! Good catch! I just now checked, and yes, the
ancient .login file (used by my prefered shell, tsch) does indeed contain:
setenv DISPLAY localhost:0.0
I probably put that in there 10+ years ago to solve some problem and then
never looked at it ever again since.
So I just simply need to remove that?? Marvelous. I'll try it.
>> I futzed around trying to fix this problem for awhile and I managed to
>> discover that if I manually set DISPLAY instead to just :0.0 then suddenly
>> all my X applications start to work again.
>Right, because it stops trying to use the TCP transport and uses local instead.
I see.
I am probably stating the obvious, but for a person like me who is not
at all steeped in the current machinations and/or machanisms of the
X server, this was/is rather entirely non-obvious.
>> Here is the output from the command "getent hosts localhost" on the system
>> in question:
>>
>> ::1 localhost
>> 127.0.0.1 localhost localhost.tristatelogic.com
>>
>>
>> What am I doing wrong? Why doesn't the default value of localhost:0.0
>> for DISPLAY work just as well as :0.0 ?
>In summary, localhost:0.0 says to use the TCP transport for the
>connection instead of local and your system is probably not listening
>on tcp.
So, just out of curiosity, when did that change?
(I must sheepishly admit that I haven't updated this system in years, so
it might even have changed 6 ofr more years ago. Obviously, I didn't get
the memo. :-)
