Re: Mouse cursor out of place?

[Fred Kiefer]

Thank you Eric for this great explanation. I had to go through all this 
myself in detail to really understand what you wrote and you are 
completely correct. It is the image that uses the wrong dpi value and 
the code in -restrictWindow:toImage: just uses the pixel size instead of 
the real image size. The problematic code is in getStandardBitmap() in 
XGServerWindow.m and it even has a comment on how to correct this.

I added a call to -setBackgroundColor: in the -init method of 
GSDragView. We will need to sort out what to do with the window size and 
the -restrictWindow:toImage: method. For the moment we cannot do without 
that method in most cases, as many users wont have compositing window 
manager, although they are becoming very popular.

Fred

On 24.06.2011 01:22, Eric Wasylishen wrote:
> Hey,
> I think I tracked down the bug..
>
> Plug.tiff had it's dpi set to 15 or so, so the point size was about 230x230. 
> The drag icon window was then set to that size (in pixels).
>
> The problem seems to be the following call, in  -[GSDragView 
> _setupWindowFor:mousePosition:imagePosition:]:
>
>    // Only display the image
>    [GSServerForWindow(_window) restrictWindow: [_window windowNumber]
>                                       toImage: [dragCell image]];
>
> which doesn't seem to work properly with images with a DPI != 72.
>
> As a quick fix you can set plug.tiff to 72dpi (using imagemagick: mogrify 
> -density 72x72 plug.tiff).
>
> a couple other notes:
> - If I do a [[self window] setBackgroundColor: [NSColor clearColor]] in 
> GSDragView, the drag icon shows up with proper alpha transparency for me - 
> since I'm using cairo and a compositing window manager.
> - _setupWindowFor:mousePosition:imagePosition: uses [image size] to set the 
> window frame - this isn't safe with resolution independence; it should set the 
> window content size instead.
> - I'm not sure what we should do to the -restrictWindow:toImage: method; 
> haven't looked at how it's implemented.
>
> Cheers,
> Eric
>
> On 2011-06-23, at 5:04 PM, Bluna Ratimonkey wrote:
>
> > Thanks Eric, and especially for letting me see it somewhere else for
> > the first time :)
> >
> >
> > BTW, do you guys think we could use something like virtual window that
> > didn't actually get mapped but blend the image with XcursorImage so
> > image doesn't have a chance to jump on load and only map the window on
> > slide back?
> >
> > On Fri, Jun 24, 2011 at 5:17 AM, Eric Wasylishen<address@hidden>  wrote:
> > > I just did a little debugging on this.. I got down to -[GSDragView 
> > > _setupWindowFor:mousePosition:imagePosition:] and the coordinates passed in to 
> > > here are correct. The frame that is set on _window in that method is correct, 
> > > but the place where the window actually shows up is at about double the 
> > > y-coordinate of where it should.
> > >
> > > I attached a screenshot. I started the drag at around (184, 180) in screen 
> > > coordinates, which is what iPoint is. You can see that _window reports it has a 
> > > frame origin at that point, but it actually appear on screen around (184, 360).
> > >
> > > -Eric
> > >
> > >
> > >
> > >
> > >
> > >
> > > On 2011-06-23, at 3:53 PM, Bluna Ratimonkey wrote:
> > >
> > > > Oh sorry,
> > > > the mouse down was
> > > >
> > > > NSPoint mouseDownPoint = [self convertPointFromBase:[anEvent locationInWindow]];
> > > >
> > > > I tried this instead without success though. Was wondering if I should do this.
> > > >
> > > >                                [self dragImage:[NSImage imageNamed:@"Plug.tiff"]
> > > >                                        at:[self convertPointFromBase:[anEvent 
> > > > locationInWindow]]
> > > >                                        offset:NSZeroSize
> > > >                                        event:anEvent
> > > >                                        pasteboard:pb
> > > >                                        source:mouseDownPort
> > > >                                        slideBack:YES];
> > > >
> > > > The case is in TMNodeView.m line 842
> > > > Thanks,
> > > > Bluna
> >
> >
> >
> > --
> > /* Join Bluna Ratimonkey (漫画家) and build the real future for GNUstep ! */
> >     [@"http://mus3.sourceforge.net"; setNeedsYourHelps:YES!!];