Xpra: Ticket #2436: after disconnect and reconnect, windows originally on different workspace appear in a single workspace.

My set up is the same as in ticket number #2434, briefly:

I am running 64-bit xpra 3.0, r24048, python3, on ubuntu

18.04 on both server and client.

The client has a dual-screen setup, with a laptop screen 1440x900 and an external display of 1280x1024. This is an extended desktop setup. I am running gnome desktop, but not wayland or gnome3 but in a "fallback" desktop environment with metacity Window manager.

I have four workspaces, in 2x2 configuration, on the client. And I put firefox and chrome (both running on the remote server and displayed via xpra) on two different workspaces on the client desktop.

After disconnecting xpra on the client, and then reconnect, all the remote window appear on one workspace, the same workspace as where I happen to be running xpra client on the command line from.

In xpra 2.* series, even after reconnect, each browser window remains in the workspace where it resided before I disconnect, which is the expected behavior.



Wed, 02 Oct 2019 10:18:41 GMT - Antoine Martin: owner changed

Can you try the python2 client / server combination? python2 /usr/bin/xpra ...

Assuming that the bug is related to python3, please attach both py2 and py3 clients -d workspace log output, otherwise one will suffice.


Fri, 04 Oct 2019 11:13:49 GMT - Jiang: attachment set


Fri, 04 Oct 2019 11:14:03 GMT - Jiang: attachment set


Fri, 04 Oct 2019 11:16:33 GMT - Jiang: attachment set


Fri, 04 Oct 2019 11:16:51 GMT - Jiang: attachment set


Fri, 04 Oct 2019 15:31:30 GMT - Antoine Martin:

Thanks for the logs, this should be fixed in r24062. Can you try to apply it by hand to see if that works for you?


Sat, 05 Oct 2019 05:33:48 GMT - Jiang:

This patch indeed helps! The window remains in their respective workspace now.

However, in this code, as in the python2 version of the 3.0 release, the window (firefox) that remains in the 2nd workspace is *a black box* after resuming from disconnect. This makes the fix actually less usable than before this patch, in python 3.0, when the client window always appear (just all in one workspace.) Is this a different bug? Perhaps I should file a separate bug report?

In any case, I attach the logs for the server and client in the patched version. perhaps this will help your debugging this.

If you want me to open another bug report, I can do that, too.


Sat, 05 Oct 2019 05:34:04 GMT - Jiang: attachment set


Sat, 05 Oct 2019 05:34:24 GMT - Jiang: attachment set


Sat, 05 Oct 2019 12:43:12 GMT - Antoine Martin:

the window (firefox) that remains in the 2nd workspace is *a black box* after resuming from disconnect

Does it remain black if you use "refresh windows" or "reinit" from the system tray menu? Does it start working properly if you move it around?


Wed, 09 Oct 2019 04:45:20 GMT - Jiang:

Sorry for the late reply again! The last few days are hectic!

Unfortunately neither "refresh window" nor "reninit" in the client side fix the black window. Moving things around does fix it, but not initially. Only after moving it to another screen (Not another desktop: remember I'm in dual screen set up), and back does it seem to revive the black window.

I've just updated to r24095 from the xpra repository, and it seems that there the black window problem persist. Should I file a new bug? For now, I prefer the old behavior. Changing workspace isn't so bad, but now I resume to a black square for one of my window.

Please let me know how I can help debugging this further.


Tue, 15 Oct 2019 14:42:21 GMT - Antoine Martin:

If you want me to open another bug report, I can do that, too.

Yes, can you please file a separate ticket? I am unclear on how you use workspaces or switch between them. Does this bug require the dual-screen or just workspaces?


Wed, 23 Oct 2019 11:35:42 GMT - Antoine Martin: status changed; resolution set

The new ticket for the black window is #2462


Sat, 23 Jan 2021 05:51:14 GMT - migration script:

this ticket has been moved to: https://github.com/Xpra-org/xpra/issues/2436