Constrain the docking direction of a window

[dimateos]

Version/Branch of Dear ImGui:

Version: 1.89.1 WIP
Branch: Docking

Back-end/Renderer/Compiler/OS: (ANY)

Back-ends: QtImGui
Compiler: MSVC_2017 / INTEL_2018
Operating System: Windows 10

My Issue/Question:

I think it would be useful to be able to limit the dock direction either horizontally or vertically for certain windows. Looking at the source (e.g. DockNodePreviewDockSetup function) I see that based on flags and more, ImGui checks that the center and the sides are available. Maybe the sides could be split into vertical and horizontal directions. I saw some relation in #3492

The rationale behind this could be, for example, a window that may be designed as a "column" with a fixed width, which requires a large vertical area to fit information displayed. Then if the user is able to dock it at the top/bottom of a dockspace most of the horizontal space is wasted and the user has to scroll. At least for dockspaces this would makes sense, for docking windows into other windows is harder to define as one could dock a "column" window on top of another. Maybe this issue can expand on the discussion.

This is also not a vital feature, but seems appropiate to reduce the user option to the one that makes sense, other UI frameworks offer it (e.g. QT as in my example). I have started recently using the docking API and I am no where close to knowing how non-trivial this could be, maybe it is not worth after all.

Thanks for the amazing work!

Screenshots/Video

Standalone, minimal, complete and verifiable example:

Open the demo example Dockspace, in Options you can set the flags for NoDockingInCentralNode and NoSplit (to disable all sides)

[ocornut]

Hello,
I likely won't be able to answer/fix that fully for a while.
Linking to #2648 which deals with the issue of docking axis (it has code but doesn't solve your problem).

[dimateos]

No problem!

It is also not a very high priority for me. I wanted to post it to suggest a possible future feature, or something to keep in mind the axis direction. Thanks for the link, if I eventually write a workaround I will post it.