Switching swipe directions should reverse app spread

This was an explicit design decision, see the following spec:

https://docs.google.com/a/canonical.com/document/d/1sNVoOXgKEOcHR5GV8VAvoDWjqGTQ1FZ-oiajuBjRt5s/edit#heading=h.yyasddj3624f

"""
Reversibility
    At which point user still has possibility to reverse back to original state? At the moment you still can reverse the gesture after first stage commit but once gone beyond second stage commit line there is no way back. You need to select an app from card view

Solved: User should be able to reverse through all the phases as long as the finger is pressed and until the second phase commit threshold is reached. After the second phase commit threshold spread is fully revealed.
"""

Truth is both the problem and solution actually seem to be the same...

The bigger right-edge spec which doesn't mention this:

https://docs.google.com/a/canonical.com/document/d/1FC_-5yz5kPy_ZvTphaqxOgI5BPnALFbxRAO4eJkzlf8/edit

I'll leave it to Vesa to comment on why that decision was taken.

For what it's worth, I had the same concern last week... the other shell gestures of launcher, greeter (horizontal too) and indicators (as well as bottom swipe) are all currently following that design choice.

Occasionally designers get it wrong :)

I suspect in this case the intent was to "click" into the spread state
rather than "sliding into more spread", and that's right once you cross
the threshold, but IF you leave your finger on the screen and reverse
direction you should "unclick" from spread and come all the way back to
the original screen if you go all the way back to the right edge.

Mark

Yeah, this is like the first implementation actually did it but design explicitly requested the "commit to spread". Will revert it back to the original behaviour. However it's probably not as simple as changing one threshold number so can't really promise it to happen for RTM unless this should be treated as a critical issue.

Vesa, can you update the specs then to reflect this when you have some time so we keep the consistency between spec and implementation?

To be clear, I do think we should "commit" in the sense that it fully
spreads once you pass the threshold. This is the "click" or "commit".
But if one does not release the finger, and reverses direction, then
passing the same point should "uncommit" to the toggle, then back to the
original screen once one gets all the way back to the right edge. In
other words, this gesture should be reversible as long as one hasn't
lifted the finger.

Mark

I remember making the call when this behaviour was designed. The reason for making the irreversible "final commit" after the second commit threshold was the big visual app movements that happen from just before the commit line position to the final spread position. I was concerned that if user was fiddling around the commit position going back and forth (i.e. with just few pixel position changes) the app screens would shoot a relatively long way back and forth on the screen.

But as far as I remember that was the only reason for the decision back then and looks like it was a wrong one. I understand the point you've made here about the reversibility and I can easily revert the decision and update the specs accordingly.

Unfortunately as Michael pointed out the implementation part is not as straight forward. But luckily I also know what you're capable of so I don't have any concerns you couldn't pull it off:) Especially since you already had it once implemented.

Yeah, we definitely want the user to have to move "back" some distance
before it "unclicks", I understand your concern and think it can be
addressed while still preserving reversibility.

Mark

review with pdMgt, not an rtm blocker

This bug was fixed in the package unity8 - 8.01+15.04.20141202-0ubuntu1

---------------
unity8 (8.01+15.04.20141202-0ubuntu1) vivid; urgency=low

  [ Michael Terry ]
  * Make sure that there is no window of opportunity for swiping away
    greeter before the lockscreen appears.
  * Don't allow dragging the greeter out from the left edge if it's
    hidden there. (LP: #1372952)

  [ Michał Sawicz ]
  * Move from Theme colour to direct UbuntuColors.orange for activity
    indicator and preview buttons

  [ Albert Astals ]
  * Use QImageReader not to load the image into memory twice (LP:
    #1384374)
  * Don't create the whole current scope delegates, just height * 3 (LP:
    #1384393, #1384374)
  * Make CroppedImageSizer async
  * Use ubuntu::connectivity::NetworkingStatus instead of
    QNetworkConfigurationManager
  * Compile with Qt 5.4 (LP: #1395636)

  [ Michael Zanetti ]
  * close previews and temp scopes on left edge/BFB (LP: #1391798)
  * make spread reversible (LP: #1368668, #1355284, #1368677)

  [ Andrea Cimitan ]
  * Different drag behaviour for not closable apps in spread (LP:
    #1368287)

  [ Daniel d'Andrada ]
  * Make UnityTestCase.touchFlick properly map event coordinates
 -- Ubuntu daily release <email address hidden> Tue, 02 Dec 2014 09:26:49 +0000

This bug was fixed in the package unity8 - 8.02+15.04.20150120~rtm-0ubuntu1

---------------
unity8 (8.02+15.04.20150120~rtm-0ubuntu1) 14.09; urgency=medium

  [ Albert Astals ]
  * Fix going to scopes when the Manage Dash is open (LP: #1403464)

  [ Josh Arenson ]
  * Create a PhysicalKeyMapper to handle all physical/hardware keys (LP:
    #1390393)

  [ Michael Zanetti ]
  * make spread reversible (LP: #1368668, #1355284, #1368677)

  [ Nick Dedekind ]
  * Unhook Lights interface from indicator widgets
 -- Ubuntu daily release <email address hidden> Tue, 20 Jan 2015 11:18:21 +0000

Thanks all!