Sorry another problem. Mouse Buttons.

[Zephro]

Ok I'm trying to tell the difference between left mouse button drops and right mouse button ones.

So I'm using the QDropEvent->mouseButtons() method to return errr well I'm not sure, how do I test these flag things exactly?

I tried if(event->mouseButtons()==Qt::leftMouseButton) which compiles fine but gives a segmentation fault.

[wysota]

The drop event is not really associated with button types. To be honest, I've never seen a "right mouse button drag".

If you want to do it anyway, try this way:


if(event->mouseButtons() & LeftButton){ qDebug("Left pressed"); }
if(event->mouseButtons() & RightButton){ qDebug("Right pressed"); }
if(event->mouseButtons() & (LeftButton|RightButton)){ qDebug("Left and right pressed"); }

[Zephro]

It's for a uni project. There's icons representing trees, if you drag them ontop of each other you get a union or conjunction based on which mouse button you're holding.

Ok I tried that, is the & a bitwise operator?

I'm still getting a segmentation fault from event->mouseButtons() which is a bit odd.... hmmmm

[wysota]

Backtrace please. With -ggdb. And the code of the event handler. Do you do some type casting there?

[Zephro]

Ok nevermind that's embaressing, the segmentation fault is because the if statements are falling through.

So both the if(event->mouseButtons() & LeftButton)
and if(event->mouseButtons() & RightButton)

are failling....


if(event->mouseButtons() & Qt::LeftButton)
{
std::cout<<"left button\n";
newTree = new TreeNode("||",this->getTree(),labelTree->getTree());

}
else
if(event ->mouseButtons() & Qt::RightButton)
{
std::cout<<"right button\n";
newTree = new TreeNode("&&",this->getTree(),labelTree->getTree());

}
else
{
std::cout<<"ignore\n";
event ->ignore();
return;
}

I'd forgotten the return statement, to stop it trying to do anything with newTree. Hmmm anyhow the if statements aren't working.

[wysota]

Probably because you are letting the buttons go to trigger the event :) Try keeping the right button pressed while releasing the left one (after causing LMB to start the drag).

[Zephro]

Aha, that works. Ta very much.

Bit counter intuitive, but it's just to demonstrate a presentation idea about trees so it's never going to be really used.

Also if you happen to know where to find the Fructherman-Reingold algorithm or Kamada-Kawai anywhere let me know! I can only find vague descriptions of the algorithms and I need something to space out the different nodes properly.

[wysota]

Aha, that works. Ta very much.

Bit counter intuitive, but it's just to demonstrate a presentation idea about trees so it's never going to be really used.


You might want to use "drop actions" instead. There are three such actions -- copy, move and link. You can use them for whatever you want (with copy being the default and two others being triggered by holding ctrl or alt). It might prove simpler.

[Chicken Blood Machine]

The drop event is not really associated with button types. To be honest, I've never seen a "right mouse button drag".

Windows Explorer is one example. It basically offers you a context menu at the drop site, so that you can choose what you intended - "copy", "move" or "link".

[wysota]

Windows Explorer is one example. It basically offers you a context menu at the drop site, so that you can choose what you intended - "copy", "move" or "link".

But you don't start the drag with RMB, right? I meant dragging with RMB more than dropping with it. Corel Draw has a neat feature -- pressing RMB during the drag modifies its type (causes making a copy of dragged object), even if you release the button before dropping the object, so it's a kind of "during drag" event. I don't know how it is done in Windows Explorer, I rather stay away from it :) I think you mean something simmilar to what KDE has -- when you drag an object to the desktop a menu pops up asking which action to take. But it is done without using RMB.

[Chicken Blood Machine]

But you don't start the drag with RMB, right?

Yes you do. (otherwise it would be very fiddly - hold left-button down; drag; press and release right-button during drag - hmmm...).


I think you mean something simmilar to what KDE has -- when you drag an object to the desktop a menu pops up asking which action to take. But it is done without using RMB.

No I don't mean that.

[wysota]

Yes you do.
I see.


(otherwise it would be very fiddly - hold left-button down; drag; press and release right-button during drag - hmmm...).

That's exactly what Corel does and it's not that bad.

[Zephro]

Ok another minor problem

It is sometimes giving this error when 2 labels merge, possibly something to do with pointers i guess as the amount of objects is changing, but I can't spot a pattern.

Anyhow the error:
ASSERT: "heartbeat != -1" in file kernel/qdnd_x11.cpp, line 910

What does that mean?

EDIT: Oh and I spent all afternoon getting QT 4.1 installed as that apparantly bug fixed something similar.

[orb9]

Also if you happen to know where to find the Fructherman-Reingold algorithm or Kamada-Kawai anywhere let me know! I can only find vague descriptions of the algorithms and I need something to space out the different nodes properly.
They are available in the boost library. But prepare for a template ride:
http://www.boost.org/libs/graph/doc/
I already tried them and did not manage to parameterize them correctly so my results where somehow useless. Do you know some good documentation for these algorithms?

[Zephro]

No idea, I just want a simple description, but detailed enough, so I can implement it myself rather than dealing with templates.

Especially as my data structures I'm working on probably won't work very well. ¬_¬