Weird TSplitter behavior

[MISV]

I added ondblclick event handler...

in that I have tried all sorts of ways to change .Left or Width of the two panels on each side.

I want the doubleclick to act as a "snap" so left side takes whole screen (and back if doubleclicked agagin)

When I doubleclick I can see it works... until... I release the mousebutton on the 2nd click in the doubleclick ... Then it all snaps back o the original placement.

My guess is that there some place inside code is some kind of management handling drags/moves which sets the position of the splitter afterwads a drag/move (and doubleclick it seems)

... Any solutions for this? Is there something like PostMessage I could use wherefrom I could set widh *after* the mouse button has been released.

I have been sitting trail and erroring this for many hours, so I figured I should hear you if you had any ideas I could try...

[skalogryz]

sample app?

[MISV]

Will make unless I find solution. Will post solution here if I find it. (Was hoping someone might have an idea I could try before I try split the UI furher to create a sample app.)

[lucamar]

Maybe add a handler for OnCanOffset? You can use OnMouseDown/Up to set a flag to know whether you're in a doubleclick and make the OnCanOffset check it and act accordingly.

Note that (at least in Linux) mouse clicks also generate OnMove events.

[MISV]

I am still working on this...

I know tried using Application.QueueAsyncCall

and then all kinds of "hacks" I can think of like disabling all event handlers while resizing using the above...

but yet... it bounces back.... So I am out of ideas almost...

It seems it sometimes when when in debug/stepping through code, but otherwise it fails completely

[skalogryz]

The first question to ask - is it LCL behavior or is it LCLCocoa behavior.

[MISV]

This seems to do the trick...

WC is the splitter

I am bit tired now so have not cut the code down the bare minimum. But here is something people can use as starting point if they suffer of the same problem:

What helped was inserting sleep before the call to the async... I think the reason is that ensures I physically actually release the mouse button (and the firing of associated messages) *before* my async code runs

Code: Pascal  [Select][+][-]

1.   P := WC.Parent;
2.   //--
3.   // Lazarus/LCL required quite a bit extra code to get this working
4.   // 1) We have disabled all event handlers to avoid more event-handler-code "firing" which --maybe helps--
5.   // 2) *Sleep* ensures we finish the physical doubleclick including release.
6.   // 3) *ProcessMessages* --maybe helps-- ensure related messages are handled
7.   // 4) *Application.QueueASyncCall* queues the rest of the code to back of messages queue
8.   // 5) *OwnHelper_Lazarus_Async_splWebsiteDblClick* also calls *ProcessMessages* as first thing which --maybe helps--  
9.   //--
10.   if WC.Left < (P.ClientWidth - WC.Width - 40) then
11.     begin
12.       {$IFDEF mymsDELPHI}
13.       W := (P.ClientWidth - WC.Width);
14.       pnlWebsiteStructuresMain.Width := W;
15.       {$ENDIF}
16.       {$IFDEF mymsFPCDevTool}
17.       Sleep(1000);
18.       Application.ProcessMessages;
19.       Application.QueueASyncCall(@OwnHelper_Lazarus_Async_splWebsiteDblClick, 1);
20.       {$ENDIF}
21.     end
22.   else
23.     begin
24.       {$IFDEF mymsDELPHI}
25.       W := Max(400, (P.ClientWidth div 4));
26.       pnlWebsiteStructuresMain.Width := W;
27.       {$ENDIF}
28.       {$IFDEF mymsFPCDevTool}
29.       Sleep(1000);
30.       Application.ProcessMessages;
31.       Application.QueueASyncCall(@OwnHelper_Lazarus_Async_splWebsiteDblClick, 0);
32.       {$ENDIF}
33.     end;
34.   ;
35.