Re: Re: Gnuradio 3.8 on Windows generates negative return codes

>  I have Visual Studio 2008 installed for C++ development. However, I do not understand how this works with python.

I'm not sure if VS 2008 supports python debugging, but I know that 2015 does, just have to be sure that the python features are installed.  Then you can create a python-based project, then add the existing .py file to it, and "Step Into" Run like a C++ file.  The only tricky part is setting up the python environment correctly, but the variables you need are all in the run_gr.bat file.

A simpler but cruder option is the old "add print 'made to here' lines everywhere" and let us know where it's dying.  Unfortunately I suspect it will be on "tb.start()" which won't provide much of a smoking gun.

Geof

On Fri, Nov 29, 2019 at 7:00 PM Lukas Haase <lukashaase@gmx.at> wrote:

Hi Geof,

Thank you very much for your help.

On 2019-11-29 17:37, Geof Nieboer wrote:
> Lukas,
>
> [...]
>
> So obviously something is failing early in the python script before it
> really gets the flowgraph going.   That error code is fairly vague,
> google tells me it's a generic system error.
>
> Please try this, open the GR command prompt, and attempt to run the
> generated python file manually, and see if it produces any more useful
> detail.

Nope, nothing:

O:\local>python test_qt.py

O:\local>

> The next step after that would be to try to run the python
> script in Visual Studio (or your preferred debugger) in debug mode and
> figure out what line is causing the failure.

Could you elaborate on this briefly?

I have Visual Studio 2008 installed for C++ development. However, I do not understand how this works with python.

> Another thing to try to would be to run it as admin and see if anything
> different occurs.  Also try setting to "No Gui" just for kicks.

As admin: exactly the same output.

No Gui: exactly the same.

> The fact that the second flowgraph does show the radio initializing is a
> good sign, it shows that block is starting up as expected.

I should have mentioned that this one is from another workstation (same version/configuration).

Additionally: Resizing the companion window makes it crash:

Assertion failed: CAIRO_REFERENCE_COUNT_HAS_REFERENCE (&surface->ref_count), file cairo-surface.c, line 955

(for this one I will file a bug report).


Thanks,
Luke