queue-developers

[Queue-developers] X11 problem

[Kai Harrekilde-P. <kh...@ex...>]

I have a slight problem with the queue-developer version.

I am unable to run Model Technology's "vsim" (a commercial VHDL
simulator) in GUI mode, while using queue.  In other words,
if I say "queue -w -i --full-pty -- vsim", vsim will run in tty mode.

I believe that I have successfully run vsim in GUI mode with
"queue -w -i --full-pty -- vsim" in the past.  Unfortunately, I
don't have any hard evidence.

Doing strace on vsim with and without queue only turns up that
stderr is not seekable when using queue.

A quick test show that "xclock" doesn't mind getting queued.

I am running queue with a local install, and identd disabled
(it gives me no end of troubles).

Any help or suggestions are welcome,


Kai
--=20
Kai Harrekilde-Petersen  <kh...@ex...>  Exbit Technology A/S
=20

Re: [Queue-developers] X11 problem

[Santeri <sa...@xp...>]

Kai Harrekilde-Petersen wrote:

> I have a slight problem with the queue-developer version.
> 
> I am unable to run Model Technology's "vsim" (a commercial VHDL
> simulator) in GUI mode, while using queue.  In other words,
> if I say "queue -w -i --full-pty -- vsim", vsim will run in tty mode.

<snip>

> Any help or suggestions are welcome,
> 
> 
> Kai

Have you tried setting DISPLAY explicitly?

Santeri

Re: [Queue-developers] X11 problem

[W. G. K. <wer...@ya...>]

The DISPLAY variable is less of a problem with Queue than xhost.

Some hosts don't need xhost set if they share a filesystem, because the X
server will store MIT_MAGIC_COOKIE authorization information in .Xauthori=
ty
on startup, which lets anyhost connect that knows the cookie. This requir=
es
that the MIT_MAGIC_COOKIE authentication scheme (insecure and outdated) b=
e
enabled on the Xserver, however.

This might be a future improvement ---

1) Automatically setting xhost on the remote host (and possibly the
propagated locate DISPLAY variable if it is not already set and there is =
a
local screen (easy))

2) Since security is a big concern with users, doing something sshd-like
(unfortunately, the code ssh is not, AFAIK, GNU but is distributed under =
a
restrictive license.) which sets up a local virtual screen on each host a=
nd
then has all X11 traffic encrypted. Since one is always allowed to connec=
t
to a local screen, this doesn't involve xhost or MIT_MAGIC_COOKIE.

Santeri Hernej=E4rvi wrote:

> Kai Harrekilde-Petersen wrote:
>
> > I have a slight problem with the queue-developer version.
> >
> > I am unable to run Model Technology's "vsim" (a commercial VHDL
> > simulator) in GUI mode, while using queue.  In other words,
> > if I say "queue -w -i --full-pty -- vsim", vsim will run in tty mode.
>
> <snip>
>
> > Any help or suggestions are welcome,
> >
> >
> > Kai
>
> Have you tried setting DISPLAY explicitly?
>
> Santeri
>
> _______________________________________________
> Queue-developers mailing list Que...@li...
> To unsubscribe, subscribe, or set options:
> http://lists.sourceforge.net/lists/listinfo/queue-developers

Re: [Queue-developers] X11 problem

[W. G. K. <wer...@ya...>]

 Lee Bradshaw <le...@se...> has informed via private email that ope=
nssh
(as opposed to the commercial F-secure product) probably has a license th=
at is
a mixture of BSD and public domain products.

I think the main thing that made openssh restrictive was the US govt's cr=
ypto
regulations, which have now (thankfully) been repealed.

However, RMS has suggested the BSD license is "restrictive" because it
requires, for instance, the "obnoxious" BSD-advertising clause. The BSD l=
icense
isn't especially compatible with GNU, although I suppose BSD is "almost"
unrestricted, so it is possible to, for example, write GNU code while
occasionally peeking at the BSD sources without too much worry about copy=
right
contamination.

If the license is a mix of public domain and BSD, it may be able to adapt=
 some
of the code (at least the public domain parts) to the GNU license.

"W. G. Krebs" wrote:

> The DISPLAY variable is less of a problem with Queue than xhost.
>
> Some hosts don't need xhost set if they share a filesystem, because the=
 X
> server will store MIT_MAGIC_COOKIE authorization information in .Xautho=
rity
> on startup, which lets anyhost connect that knows the cookie. This requ=
ires
> that the MIT_MAGIC_COOKIE authentication scheme (insecure and outdated)=
 be
> enabled on the Xserver, however.
>
> This might be a future improvement ---
>
> 1) Automatically setting xhost on the remote host (and possibly the
> propagated locate DISPLAY variable if it is not already set and there i=
s a
> local screen (easy))
>
> 2) Since security is a big concern with users, doing something sshd-lik=
e
> (unfortunately, the code ssh is not, AFAIK, GNU but is distributed unde=
r a
> restrictive license.) which sets up a local virtual screen on each host=
 and
> then has all X11 traffic encrypted. Since one is always allowed to conn=
ect
> to a local screen, this doesn't involve xhost or MIT_MAGIC_COOKIE.
>
> Santeri Hernej=E4rvi wrote:
>
> > Kai Harrekilde-Petersen wrote:
> >
> > > I have a slight problem with the queue-developer version.
> > >
> > > I am unable to run Model Technology's "vsim" (a commercial VHDL
> > > simulator) in GUI mode, while using queue.  In other words,
> > > if I say "queue -w -i --full-pty -- vsim", vsim will run in tty mod=
e.
> >
> > <snip>
> >
> > > Any help or suggestions are welcome,
> > >
> > >
> > > Kai
> >
> > Have you tried setting DISPLAY explicitly?
> >
> > Santeri
> >
> > _______________________________________________
> > Queue-developers mailing list Que...@li...
> > To unsubscribe, subscribe, or set options:
> > http://lists.sourceforge.net/lists/listinfo/queue-developers
>
> _______________________________________________
> Queue-developers mailing list Que...@li...
> To unsubscribe, subscribe, or set options:
> http://lists.sourceforge.net/lists/listinfo/queue-developers

Re: [Queue-developers] X11 problem

[QingLong <qin...@Bo...>]

On Fri, Mar 23, 2001 at 11:38:07AM +0100, Kai Harrekilde-Petersen wrote:
>
> I have a slight problem with the queue-developer version.
>
   Excuse me if I have missed it, but
 what version did you use? Did you try to run it (queued and queue)
 with --verbose and/or --debug flags? Did you notice something abnormal
 in those logs?

>
> I am unable to run Model Technology's "vsim" (a commercial VHDL
> simulator) in GUI mode, while using queue.  In other words,
> if I say "queue -w -i --full-pty -- vsim", vsim will run in tty mode.
>
   Have `vsim' got some command line option to force it running in tty mode,
 X-window mode? If it has no such flag, then how does it decide what mode
 to run in? Does it just look for DEISPLAY set? Or some private magic?

>
> I believe that I have successfully run vsim in GUI mode with
> "queue -w -i --full-pty -- vsim" in the past.  Unfortunately, I
> don't have any hard evidence.
>
   Can you recall the version of that queue that had managed to do it right?

>
> Doing strace on vsim with and without queue only turns up that
> stderr is not seekable when using queue.
>
   And how about stdout? Is it seekable? Hmmm, I believe it shouldn't be.

      QingLong.