Window management - Change spread ordering from recency to spatial

[John Lea]

By default, windows are shown in a order that corresponds as closely as possible to the physical location of the windows on the screen (spatial layout). If two or more windows are in the exact same location (e.g. both windows are maximised), they should be shown in the order in which they were created.

For more details see section "2.1. Application Spread" in https://docs.google.com/a/canonical.com/document/d/1EdrlUuZvA9P8-BZufUU2KlHGjg49p9UacF4MCL0U5uA/edit#heading=h.pgfs08v5wuv3

Visual designs are available at http://design.canonical.com/the-toolkit/unity-task-switching/

[Sam Spilsbury (smspillaz)]

Difficulty: Easy
Method: Test and make sure this branch still works: https://code.launchpad.net/~smspillaz/unity/unity.scale_layout
Test: Need to get the window layout code under unit test. Talk to jason about it.

[Andrea Azzarone (azzar1)]

IIRC we already show windows «in a order that corresponds as closely as possible to the physical location of the windows on the screen».

[Olivier Tilloy (osomon)]

That’s correct, the scale plugin computes slots so as to minimize the distances between the windows’ original position and the corresponding slots.

Need to check whether recency is used as a criterion to decide between two windows that have the exact same position.

[Olivier Tilloy (osomon)]

I just tested the feature with four firefox windows and the computation of the slots according to the original position of the windows seems to work. With four windows, I’m seeing two rows, the first one with three thumbnails and the second one with only one. This looks odd to me, I would have expected two thumbnails on each row, but in any case that’s a separate issue.

However when all four windows have the same position, they are not ordered by recency, so that needs fixing.

[Sam Spilsbury (smspillaz)]

Right, the sorting order is based on recency. You'll need to tweak the layout algorithm to take that into account the window position as an option.

[Olivier Tilloy (osomon)]

@Sam: your last comment got me confused.

From a quick glance at the code, it looks to me like the current algorithm to position the windows in the spread minimizes the distance to the slot. So that would be the first half of the design specification already implemented. What seems to be missing is the use of the recency criterion to decide between two windows that have the exact same geometry.

Or am I misunderstanding the current state of the code?

[Andrea Azzarone (azzar1)]

Should be already fixed.