allow the client more control over window update speed

[totaam]

Issue migrated from trac ticket # 543

component: core | priority: major | resolution: fixed

2014-03-20 11:24:47: totaam created the issue

---

The client may offer an enhanced UI to let the user select which windows are more important (ie: #539), or we can detect which windows are more important: if a window is not shown on the current desktop, we don't need to refresh it at full speed!

[totaam]

2014-03-20 11:29:32: totaam uploaded file workspace-change-batch-delay.patch (16.8 KiB)

when the window is not on the current workspace, limit its update speed

[totaam]

2014-03-20 11:31:59: totaam changed status from new to assigned

[totaam]

2014-03-20 11:31:59: totaam changed owner from antoine to totaam

[totaam]

2014-03-20 11:31:59: totaam commented

---

The patch above is in pretty good shape, it overloads the "buffer-refresh" packet with the ability to:

• update the batch configuration fairly generically
• update "client_properties" so we don't need to use a "configure-window" packet for property changes

The window refresh drops dramatically when it isn't on the current desktop (though still higher than I expected - to investigate)

[totaam]

2014-03-20 14:11:20: totaam uploaded file window-throttled.png (13.7 KiB)

before moving the window to another workspace on the left, after on the right

[totaam]

2014-03-20 14:13:39: totaam commented

---

Merged in r5870, it works really well.

Here's the most extreme example, using the speed setting "low latency" with x264:
[[Image(window-throttled.png)]]

With more reasonable settings, the bandwidth savings are still impressive, usually about 8 times less once the throttling kicks in.

I am keeping the ticket open for further testing, and probably refactor the code a little to use it on suspend / resume (see #492)

[totaam]

2014-03-21 04:32:25: totaam changed status from assigned to closed

[totaam]

2014-03-21 04:32:25: totaam changed resolution from ** to fixed

[totaam]

2014-03-21 04:32:25: totaam commented

---

Hooked into the power events callbacks in r5875 - see #492.

Will probably do more on this as part of #540.