layouts branch has been merged into devel

[Edward K. Ream]

Details are in PR 1691. No significant new capabilities are present, but the code is simpler and I think it is worth doing.

Please report any new layout-related problems immediately.

Edward

[Thomas Passin]

I do have a problem.  It's with add_adjacent(), in class NestedSplitter.  It has an inner function hunter(layout, id_).  hunter() expects to receive a layout dictionary with a key 'content'. However , when add_adjacent() is called by VR3, what gets handed to hunter() is not a dictionary but a string that appears to be a string representation that the value of the content key would normally have - at least, if the string were split(), it would look like that.  So the hunter() call fails.

I tried changing the input parameter to the implied dictionary (by modifying the code in nested_splitter.py), but then it got into an unending series of recursive calls.

So something is amiss.

[Edward K. Ream]

On Sun, Oct 11, 2020 at 9:20 PM Thomas Passin <tbp1...@gmail.com> wrote:

I do have a problem.  It's with add_adjacent(), in class NestedSplitter.  It has an inner function hunter(layout, id_).  hunter() expects to receive a layout dictionary with a key 'content'. However , when add_adjacent() is called by VR3, what gets handed to hunter() is not a dictionary but a string that appears to be a string representation that the value of the content key would normally have - at least, if the string were split(), it would look like that.  So the hunter() call fails.

I see this too. I'll fix it today.

Edward

[Edward K. Ream]

The fix is in the layout2 branch. The corresponding PR is #1705. I have walked back some of the changes I made to Terry's code, but the actual cause of the crash is mysterious.

Please test the layout2 branch and report any problems. Unlike before, I have tested the code with both the VR and VR3 plugins with no apparent problems.

Edward

[Thomas Passin]

On Monday, October 12, 2020 at 12:09:24 PM UTC-4, Edward K. Ream wrote:

The fix is in the layout2 branch. The corresponding PR is #1705. I have walked back some of the changes I made to Terry's code, but the actual cause of the crash is mysterious.

Please test the layout2 branch and report any problems. Unlike before, I have tested the code with both the VR and VR3 plugins with no apparent problems.

The first time I opened VR3, it opened in one of those titlebar-off-screen floating windows.  When I tried to toggle it off, Leo spewed out a continuous stream of the following:

qt.qpa.backingstore: composeAndFlush: makeCurrent() failed

After I closed and reopened Leo, I used the new command vr3-use-default-layout and then VR3 opened normally in the right-hand panel. After that it continued to open and close without errors.

Note that this was using the version of VR3 in the layout2 branch.  Using my own slightly modified version, VR3 opened and closed correctly, including when a floating window was opened using the splitter context menu.

[Thomas Passin]

I made a few changes to my own version, and they eliminated an error I got when I tried to invoke the new vr3-use-default-layout command after opening VR3 using the splitter context menu Insert/viewrendered3.

On Monday, October 12, 2020 at 1:32:46 PM UTC-4, Thomas Passin wrote:

Note that this was using the version of VR3 in the layout2 branch.  Using my own slightly modified version, VR3 opened and closed correctly, including when a floating window was opened using the splitter context menu.

I'll do a PR for the modified code into the layout2 branch.