Debugger unresponsive in RunUserInterface

[Guillaume Dargaud]

Hmmm,
I have a strange problem this morning: my program locks up as soon as it
enters RunUserInterface. The panels are unresponsive and don't redraw. I
can't do anything in the UIR. I can't even break the program in the
debugger...
And I haven't changed much since yesterday !
Any hint ?
That's in CVI 8.5.1 on Windows.
--
Guillaume Dargaud
http://www.gdargaud.net/

[Guillaume Dargaud]

And the global break key just kills the IDE and the program... C:-(
--
Guillaume Dargaud
http://www.gdargaud.net/

[Guillaume Dargaud]

OK, I _think_ I have found a bug in the debugger...
I don't have time to make a CVI project that exhibits the problem (sorry, I
have a deadline today), but here's the jest of it:

int Val; // May not be necessary

void Function(...) {
char Val[255]; // May not be necessary
....
{ // Some free block
double Val; // Place a watch on this
....
}

} // End Function

Try to place a watch on the innermost Val, in some/most cases you can't, the
debugger won't let you do it.
Furthermore, if you manage to place a watch on it, the program deadlocks
when entering RunUserInterface.

There may be other issues at play such as threads, but the thing is, if I
have a watch on a free-block variable, everything locks up.
--
Guillaume Dargaud
http://www.gdargaud.net/
"They don't make bugs like Bunny anymore." Olav Mjelde.

[nickb]

Hello Guillaume,

Thanks for reporting this.  I have confirmed a bug with the pseudo code you have posted, and filed a Bug Report (#155604) for further investigation.  We'll let you know as we have more information.

NickB

National Instruments 

[dummy_decoy]

ha, this seems to be a new behavior: on older versions of CVI (prior to 8.5), the debugger would just ignore the variable inside the free block, complaining that the symbol does not exist. since this time, i gave up placing a watch on a variable inside a free block...

[Guillaume Dargaud]

> Hello Guillaume, Thanks for reporting this.  I have confirmed a
> bug with
> the pseudo code you have posted, and filed a Bug Report (#155604) for
> further
> investigation.  We'll let you know as we have more
> information. NickBNational Instruments 

OK, I wasn't going crazy then... good to know !

And while I'm at it... Here's another bug:
- fold some code block in the editor (click on the [-])
- press Ctrl-Y to remove the line (where the entire block is now hidden)
- funny things happen... C;-)
--
Guillaume Dargaud
http://www.gdargaud.net/

[LuisG]

Hello Guillaume,

Thanks for reporting this. I was able to reproduce this bug, I think. But I want to run the symptom I see by you, just to make sure that I'm not missing some additional ones:

1. It only deletes the visible line, not the entire collapsed block

2. The cursor moves temporarily to the top of the window, to an invalid location, and then moves back down to the correct location.

3. The contents of the window won't redraw correctly until a few seconds have ellapsed.

Are you seeing any other symptoms?

By the way, if you highlight the line and then remove it using <Ctrl-X> (which is the same as <Ctrl-Y>) the problem doesn't happen.

I've recorded the bug with ID #156166.

Thanks,

Luis

 Read more >>   Options >>  

[LuisG]

Guillaume and dummy_decoy,

On the original issue that Guillaume reported, I just wanted to clarify that CVI does not support using block-scope variables in watch points. In the example provided, you are of course able to use the outer Val variable (the string array) in a watchpoint, and you are then able to interpret this variable as a double, if you so choose. But this reinterpretation doesn't mean that the watch point will then be using the other variable. It just has the effect of changing the type of the original string array variable. The value that it would then display would be garbage.

Having said, the hang that Guillaume saw is still unexpected, and we still need to investigate its cause.

Luis

[Guillaume Dargaud]

Yes, I see the same symptoms as you.
It's a minor bug, but still a bit unsettling when it happens.
--
Guillaume Dargaud
http://www.gdargaud.net/