Re: [Rosegarden-devel] misc , backtraces

[Emanuel R. <em....@go...>]

There were troubles with my mail account, I didn't notice soon.
I didn't get any mail the 31.10. and 01.11.

I'm trying to catch up with the archive now.



>
> From: D. Michael McIntyre <michael.mcintyre@ro...> - 2009-10-31 01:04
> On Friday 30 October 2009, Emanuel Rumpf wrote:
>
>> But one has to know, that the controllers are bound to the device as well.
>
> Well, they *are* bound to the device, so there's no other way to do it. These
> things are very highly variable, and very highly device-specific.

On the other Hand: for the interpreter (e.g. the softsynth) it doesn't
matter, where
the controller 0x08 data comes from.

Also it's quite possible, that one has many devices which can send MidiCCs.
For every device I would have to setup the CCs again.

What I want do say: With some (custom) CCs assigned, it is complicated
to switch the device (with the way it is currently working).

>
>> Also I had to manually choose an unused position nr. for the controller,
>> if not, then it showed a wrong name in the IPB.
>
> I don't like that whole picking a number to choose a position thing. It's
> garbage. If someone wants a little project, come up with some kind of little
> WYSIWYG controller-arranger widget where you can drag things around and show
> and hide them through a nice usable interface instead of this tripe.
>
A good idea.

> As far as that goes though, you shouldn't actually have to have your new
> controller showing for it to work on a control ruler.
>
>> I don't believe this theory. If it crashes (any way), there must be
>> something more wrong with it.
>
> Chris Cannam could explain it better than me. I don't remember the details,
> but there is no doubt in my mind that we want to "delete on close" these
> things. If you want something to be persistent, use QSettings.
> --
> D. Michael McIntyre
>



> From: D. Michael McIntyre <michael.mcintyre@ro...> - 2009-10-31 01:10
> On Friday 30 October 2009, Emanuel Rumpf wrote:
>> It works. (After disabling some code in ColourMap.cpp )
>
> What code? Your knobs are all black.
>
I disabled
getColourByIndex()
and
getColourByName()
which caused a crash, when working with the rulers,
thus the knobs are black on the screenshot.


>> Here is a screenshot with midi controller 08 recorded.
>> But it is clipped at top and bottom, meaning not the full range
>> (0...127) is visible ?
>
> Forget the real problems! :) Why do you have a red background in so many
> places? Did you customize your stylesheet?
>
Yes. I did change the background color.
While most of the stylework you did is great, the black background
looks horrible to me. :-)


> I'll look into the crashes on this. I had been operating under the assumption
> that the ColourMap crashes were studio-related, but maybe there's more to it.
>
I don't know where they come from. Maybe it isn't properly initialized?
>
> --
> D. Michael McIntyre
>


>
> From: D. Michael McIntyre <michael.mcintyre@ro...> - 2009-10-31 02:37
> On Friday 30 October 2009, D. Michael McIntyre wrote:
>
>> On Friday 30 October 2009, Emanuel Rumpf wrote:
>> > It works. (After disabling some code in ColourMap.cpp )
>>
>> What code? Your knobs are all black.
>
> I think I figured it out. You disabled code in ColourMap to stop it from
> crashing, but that wasn't the real reason why it was crashing.
>
> This turned out to be another example of why we
>
> setAttribute(Qt::WA_DeleteOnClose);
>
> What was happening, I'd open a ControlEditorDialog, add some controllers and
> do my thing, close it, close the device manager, and everything would be fine
> until I drew something with the segment pencil.
>
> Then ControlEditorDialog would crash in slotUpdate() even after it was closed.
> Why was a closed dialog trying to update itself anyway? I'm not sure, but I
> imagine it must have caught a signal from something.
>
> Add the delete on close, and the problem goes away as far as I can tell. No
> more crash, and I don't see any strange side effects.
>
DeleteOnClose for the ControlEditorDialog ? Ok, If it's not in svn
already, I will try.


--
E.R.