[regression] Extra flickable margin added when using PullToRefresh

The bug was not introduced by the MR linked in the bug description. I reproduced the bug with this code, and pulling down the listview and releasing it with a vertical velocity:

import QtQuick 2.2
import QtQuick.XmlListModel 2.0
import Ubuntu.Components 1.3
import Ubuntu.Components.ListItems 1.3
MainView {
    width: units.gu(40)
    height: units.gu(71)
    Page {
        id: page0
        title: "Reuters"
        ListView {
            id: view
            anchors.fill: parent
            model: 15
            delegate: Standard {
                width: ListView.view.width
                height: units.gu(5)
                text: index//title
            }
            onVerticalVelocityChanged: print("vv = "+verticalVelocity)
            PullToRefresh {
                id: pullToRefresh
                onRefreshingChanged: print("refreshing = "+refreshing)
                onRefresh: {
                    refreshing = true;
                    refreshing = false;
                    // refreshTimer.start();
                }
            }
        }
    }

// Timer {
// id: refreshTimer
// onTriggered: pullToRefresh.refreshing = false
// }
}

and with UITK r1930 (before the mentioned MR) and r1918.

I think the bug was always there, the topMargin gets messed up by the PullToRefresh. However, pre-r1931, the topMargin was (incorrectly) reset to 0 when a new Page was pushed on the PageStack (which happens in onRefresh for the example linked to th ebug). Now that this no longer happens, the bug becomes visible when the Page is popped from the stack.

I reproduced the bug with this code, which has no header:

import QtQuick 2.2
import QtQuick.XmlListModel 2.0
import Ubuntu.Components 1.3
import Ubuntu.Components.ListItems 1.3
Item {
    width: units.gu(40)
    height: units.gu(71)
        id: page0
        ListView {
            id: view
            anchors.fill: parent
            model: 15
            delegate: Standard {
                width: ListView.view.width
                height: units.gu(5)
                text: index
            }
            onVerticalVelocityChanged: print("vv = "+verticalVelocity)
            PullToRefresh {
                id: pullToRefresh
                onRefreshingChanged: print("refreshing = "+refreshing)
                onRefresh: {
                    refreshing = true;
                    refreshing = false;
                }
            }
        }
    }

To summarize, there were 2 bugs:

1. PullToRefresh messes up the topMargin
2. On pop(), the topMargin of page0 was set to header.height (instead of keeping the current topMargin) (Fixed)

Bug 2 was hiding bug 1 in the example from the report, and it was exposed when bug 2 got fixed. We still need to fix bug 1, but a workaround for now is to reset the incorrect topMargin when popping a page.

Note that the example in my previous comment shows bug 1 independently from bug 2.

If the refreshing takes some time, as in http://paste.ubuntu.com/16318581/ then I cannot reproduce the bug on my laptop. In the linked example, the interval is 200 ms. If I reduce that to <100ms, then I can reproduce the bug.

This bug was fixed in the package ubuntu-ui-toolkit - 1.3.1984+16.10.20160527.2

---------------
ubuntu-ui-toolkit (1.3.1984+16.10.20160527.2) yakkety; urgency=medium

  [ Christian Dywan ]
  * Slimmer frame for TextFields: 0.5dp. Fixes LP: #1578190.

  [ Albert Astals Cid ]
  * Add override
    The override specifier (since C++11) specifies that a virtual function
    overrides another virtual function. In a member function declaration or
    definition, override ensures that the function is virtual and is overriding
    a virtual function from the base class.

  [ Tim Peeters ]
  * Fix reference error in PullToRefreshStyle. Fixes LP: #1582843
  * Mark Tab, Tabs, TabBar, PageHeadConfiguration, PageHeadSections,
    PageHeadState, ToolbarButton, ToolbarItems as deprecated in the
    documentation. Fixes LP: #1566735, LP: #1566741.

  [ CI Train Bot ]
  * Resync trunk.

 -- Zoltan Balogh <email address hidden> Fri, 27 May 2016 07:08:44 +0000

I just tried the test application I attached to the bug, and I can still reproduce this.