[Interest] Assertions from QQuickTableView::forceLayout()

[Patrick Stinson]

TableView.forceLayout() seems prone to throwing assertions related to internal data structures being out of sync with the number of rows and columns from the QAbstractTableModel used. Some of these assertions were fixed in 5.13.0, more were fixed in 5.13.1. But this one still remains:

items/qquicktableview.cpp:458:void QQuickTableViewPrivate::dumpTable() const(): "table cells: (4,0) -> (15,15), item count: 160, table rect: 270,0 x 1040,704"
items/qquicktableview.cpp:477:auto QQuickTableViewPrivate::modelIndexAtCell(const QPoint &)::(anonymous class)::operator()() const(): output: modelIndex: 915 cell: QPoint(15,0) count: 915
items/qquicktableview.cpp:477:None(): ASSERT: "(modelIndex < model->count()) || [&](){ dumpTable(); qWarning() << "output:" << "modelIndex:" << modelIndex << "cell:" << cell << "count:" << model->count(); return false;}()" in file items/qquicktableview.cpp, line 477

In many cases this assertion happens when TableView.rows is out of sync with model.rowCount(), or TableView.columns is out of sync with model.columnCount(). The fact that these are out of sync at all seems to me to be a bug?

At any rate, this isn’t a reliable workaround. This happens most often when emitting columnsRemoved() from the model while the column is visible. It never occurs when the column is hidden, e.g. scrolled beyond the width of the TableView’s viewport.

Is there something I’ve missed about the TableView implementation? Or does it just early days yet for this QtQuick Item?

And for what it’s worth, QQuickTableView::dumpTable() was also throwing an assertion regarding the GL context not matching when dumping the table to a png. I commented out that code block to be able to see the original assertion above.

Cheers,
-P
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.qt-project.org/pipermail/interest/attachments/20190920/cf6251d9/attachment.html>