Re: Running tests with xvfb

2017-08-02 Thread Jeff
On 29/07/17 12:05, Christian Seiler wrote:
> Is that actually true? Because xvfb should not interact with any
> regular X sesssion that is running.

Now you are right. I can't reproduce the problem any more.

> What is the value of the DISPLAY environment variable the program
> is trying to use? Is that the virtual display created by Xvfb, or
> is it the actual X session you're running on?

Inside xvfb it is now :99

Before I tried all sorts of things to get it to ignore :0.

Thanks for all the pointers.

Dunno what was going on.

Regards

Jeff



signature.asc
Description: OpenPGP digital signature


Re: Running tests with xvfb

2017-07-31 Thread Ian Jackson
Jeff writes ("Running tests with xvfb"):
> I have a package whose tests crash X on my machine, which uses nouveau.
> This makes testing rather inconvenient.
> 
> Running the tests in a chroot with xvfb works, but takes an age (i.e. a
> couple of minutes) to set up the chroot. This is also not conducive to
> rapid testing of small changes.

1. Why are you setting up the chroot each time ?  You could keep it
set up.

> Running the test outside the chroot with xvfb still crashes X, because
> xvfb seems to grab the "real" X if it is there.

2. I think you mean "the test grabs the real X" not that xvfb grabs
it.

3. Probably the actual difference is that when you run the tests in a
chroot, something launders your environment so that DISPLAY is not
set.  (Or maybe that your chroot tool is providing a separate AF_UNIX
display numbering space, so that a default of ":0" now goes to Xvfb.)

If I am right you can fix this problem by starting Xvfb in your main
session, and running the tests with the appropriate setting of
DISPLAY.

If this does not work then I think it is a bug in the tests, which
should honour DISPLAY.

Ian.



Re: Running tests with xvfb

2017-07-29 Thread Julien Cristau
On Fri, Jul 28, 2017 at 22:46:57 +0200, Jeff wrote:

> Running the test outside the chroot with xvfb still crashes X, because
> xvfb seems to grab the "real" X if it is there.
> 
> Is there a way of getting xvfb to ignore the system X?
> 
Xvfb doesn't do anything with your regular X server so that can't be
right.

Cheers,
Julien



Re: Running tests with xvfb

2017-07-29 Thread Paul Wise
On Sat, Jul 29, 2017 at 5:45 AM, Jeff wrote:

> I did that six months ago:
>
> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=857620

Looks like that received no responses, you might want to re-test with
the latest Xorg in sid/experimental and forward the results to nouveau
upstream.

-- 
bye,
pabs

https://wiki.debian.org/PaulWise



Re: Running tests with xvfb

2017-07-29 Thread Christian Seiler
Hi,

On 07/28/2017 10:46 PM, Jeff wrote:
> I have a package whose tests crash X on my machine, which uses nouveau.
> This makes testing rather inconvenient.
> 
> Running the tests in a chroot with xvfb works, but takes an age (i.e. a
> couple of minutes) to set up the chroot. This is also not conducive to
> rapid testing of small changes.
> 
> Running the test outside the chroot with xvfb still crashes X, because
> xvfb seems to grab the "real" X if it is there.
> 
> Is there a way of getting xvfb to ignore the system X?

Is that actually true? Because xvfb should not interact with any
regular X sesssion that is running.

Could not the more likely thing be that while you start xvfb the
program you're trying to test uses your regular X session instead
of connecting to xvfb?

What is the value of the DISPLAY environment variable the program
is trying to use? Is that the virtual display created by Xvfb, or
is it the actual X session you're running on?

Regards,
Christian



Re: Running tests with xvfb

2017-07-29 Thread Jeff
On 28/07/17 23:19, Eric Cooper wrote:
> Can you use an xorg.conf with the "dummy" driver instead of xvfb?
> I use a "with-dummy-xserver" wrapper script for situations like that.

Can I do this in parallel with my standard display driver, so that the
dummy is only used when I call it from the wrapper?

Regards

Jeff



signature.asc
Description: OpenPGP digital signature


Re: Running tests with xvfb

2017-07-29 Thread Jeff
On 29/07/17 03:31, Paul Wise wrote:
> This sounds like an Xorg/nouveau bug, please do report it:
> 
> http://x.debian.net/howto/report-bugs.html

I did that six months ago:

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=857620

Regards

Jeff



signature.asc
Description: OpenPGP digital signature


Re: Running tests with xvfb

2017-07-28 Thread Paul Wise
On Fri, Jul 28, 2017 at 4:46 PM, Jeff wrote:

> I have a package whose tests crash X on my machine, which uses nouveau.

This sounds like an Xorg/nouveau bug, please do report it:

http://x.debian.net/howto/report-bugs.html

-- 
bye,
pabs

https://wiki.debian.org/PaulWise



Re: Running tests with xvfb

2017-07-28 Thread Eric Cooper
On Fri, Jul 28, 2017 at 10:46:57PM +0200, Jeff wrote:
> I have a package whose tests crash X on my machine, which uses nouveau.
> This makes testing rather inconvenient.
>
> Running the tests in a chroot with xvfb works, but takes an age (i.e. a
> couple of minutes) to set up the chroot. This is also not conducive to
> rapid testing of small changes.
>
> Running the test outside the chroot with xvfb still crashes X, because
> xvfb seems to grab the "real" X if it is there.
>
> Is there a way of getting xvfb to ignore the system X?

Can you use an xorg.conf with the "dummy" driver instead of xvfb?
I use a "with-dummy-xserver" wrapper script for situations like that.

--
Eric Cooper e c c @ c m u . e d u



Running tests with xvfb

2017-07-28 Thread Jeff
Hallo all,

I have a package whose tests crash X on my machine, which uses nouveau.
This makes testing rather inconvenient.

Running the tests in a chroot with xvfb works, but takes an age (i.e. a
couple of minutes) to set up the chroot. This is also not conducive to
rapid testing of small changes.

Running the test outside the chroot with xvfb still crashes X, because
xvfb seems to grab the "real" X if it is there.

Is there a way of getting xvfb to ignore the system X?

Regards

Jeff



signature.asc
Description: OpenPGP digital signature