I've a problem with my Qt/interview application. I use QTreeView to display tree data. I implemented my own model based on QAbstractItemModel.
I get a following error prior to application crash. It happens often after I add new record.
Could You explain to me what is the meaning of this error. What is a QPersistentModelIndex ?
I'm not using QPersistentModelIndex in my code.
ASSERT failure in QPersistentModelIndex::~QPersistentModelIndex: "persistent model indexes corrupted"
Thanks.
QPersistentModelIndexes are (row, column, parent) references to items that are automatically updated when the referenced items are moved inside the model, unlike regular QModelIndex. For instance, if you insert one row, all existing persistent indexes positioned below the insertion point will have their row property incremented by one.
You may not use them directly, but QTreeView does, to keep track of expanded items and selected items, for example.
And for these persistent indexes to be updated, you have to call the functions QAbstractitemModel::beginInsertRows() and endInsertRows() around the actual row insertion(s) when you add new records.
See the end of the section about subclassing model classes for details: http://doc.trolltech.com/latest/qabstractitemmodel.html#subclassing
I found this method QAbstractItemModel::persistentIndexList and I'm
wondering what indexes it should return. All of them ?
Should this method return all nodes currently visible in the TreeView ?
That method returns only the indexes for which a QPersistentIndexModel was created and is still in scope (as a local variable, a class member, or in a QList<QPersistentIndexModel> for example).
Expanded or selected nodes are not necessarily currently visible, so you can't (and shouldn't anyway) assume anything about what these persistent indexes are used for.
You just have to keep them updated, and you only need to use persistentIndexList for big changes in the model, like sorting (see QTreeWidget internal model : QTreeModel::ensureSorted(link)), for smaller incremental changes you have all the beginXxxRows/beginXxxColumns and endXxxRows/endXxxColumns methods.
Related
I want to know how to change the order of QTableView rows by dragging, and store the order changes to the model?
I use QTableView as view and QSqlTableModel as model. I am using Qt 5.15.
I set:
ui->table_view->setSelectionMode(QAbstractItemView::SingleSelection);
ui->table_view->setSelectionBehavior(QAbstractItemView::SelectRows);
ui->table_view->setEditTriggers(QAbstractItemView::NoEditTriggers);
Methods I tried:
1.
ui->table_view->setSectionsMovable(true);
ui->table_view->setDragEnabled(true);
ui->table_view->setDragDropMode(QAbstractItemView::InternalMove);
ui->table_view->setAcceptDrops(true);
It doesn't take effect. The row cannot be dragged.
2.
ui->table_view->verticalHeader()->setSectionsMovable(true);
ui->table_view->verticalHeader()->setDragEnabled(true);
ui->table_view->verticalHeader()->setDragDropMode(QAbstractItemView::InternalMove);
ui->table_view->verticalHeader()->setAcceptDrops(true);
The row can be dragged by vertical header. But the order of changes will not affect the model.
As far as I know the QTableView does not call moveRows method of the model by itself. Instead drag and drop actions call *mimeData methods.
So, one option would be the following:
In your model you need to implement mimeData and dropMimeData functions. When you drag and drop a row the view will ask for mimeData() of the given row. If you are sure that you need only internal moves you can encode the selected row indices to QMimeData. Then in the dropMimeData() you decode the indices that were selected and use them to call your moveRows() implementation. Return false from this function to prevent removing of the moved out rows.
Another option can be to override the QTableView methods such as dropEvent() in a way, that it calls the model moveRows() method directly.
My friends, Could you please explain the concept of these methods to me? Normally, when I get a data I would use 'index' to refer to items in a model and specify roles via data attribute. I came across 'item' method today and the explanation of this method is "Returns the item for the given row and column if one has been set; otherwise returns 0." What are differences from 'index'+'data' method? Is it just a shortcut?
This is the documentation of 'item' method.
https://doc.qt.io/qt-5/qstandarditemmodel.html#item
Alternatively, this is the documentation of 'index' method.
https://doc.qt.io/qt-5/qstandarditemmodel.html#index
I'll try my best to explain it.
The item is like the actual widget you see in the view (it's not actually a widget, but I think it's a good way to think about it). It's what the user is actually seeing.
The index is more "behind the scenes." It's like a pointer to a position in the model.
An item can exist without an index. But a valid index cannot exist without an item. The item only associates with an index when it is put into a model. Otherwise, it's just an item that no one can look at.
Take an array as an example... It contains multiple "items". You specify which item from the array you want by providing a number, aka the index. Simply put, the index only exists when it is associated with an item in the array. But the item can exist outside of the array and be it's own thing without an index.
The QModelIndex was created to be a lightweight way to reference items in a model. Similar to the way you can use a number to represent an object stored in an array.
I'm new to RxJava, and trying to use the Realm Observable feature.
When doing this
realm.where(Intake.class)
.findAllSorted("time", Sort.DESCENDING)
.asObservable()
I get the full list of items, but when soemthing changes (ie item added), I get the full list again.
What is the RxJava-way to get only the new items?
Thanks in advance
You are currently observing the query results. Because your query can contain multiple items (findAllSorted()), you're always observing the RealmResults that will emit all sorted items when there's a change (see docs).
You can do something like this:
realm.where(Intake.class)
.findAllSorted("time", Sort.DESCENDING)
.asObservable()
.flatMapIterable(results -> results)
.distinct();
This does 2 more things:
convert your RealmResults to singular Intake instances
only let distinct items pass through (make sure your Intake implements equals() correctly)
This does however impose some extra CPU load, because each time the query passes on a new RealmResult, processing is done to filter out the distinct items.
In the above example, sorting will work on the initial set of emitted Intake objects. However, any subsequent emitted items could be observed out of order because they are new and emitted after the initial results.
I'm displaying the query from an QSqlModel in a table view. e.G.
SELECT id, name FROM person;
I don't want to display id in the view, however I need the id when the user clicks on a row in that view.
Is there a model in Qt, which allows me to display only specified columns in views but also providing access to invisible ones?
I used to do it by hiding the columns e.g.:
ui->myview->setColumnHidden( 0, true );
But probably there is a better solution.
Since the OP amended the question to exclude using QTableView the only alternative that would seem to work is the creation of a new class derived from QSortFilterProxyModel. One can override the filterAcceptsColumn method
Documentation on the class QSortFilterProxyModel can be found here
QSortFilterProxyModel can be used for sorting items, filtering out items, or both. The model transforms the structure of a source model by mapping the model indexes it supplies to new indexes, corresponding to different locations, for views to use. This approach allows a given source model to be restructured as far as views are concerned without requiring any transformations on the underlying data, and without duplicating the data in memory.
You'd also have to override MapToSource and SourceToMap . There is a good example of usage in this StackOverflow's answer
I am managing a large dateset using crossfilter in my recent project, in which I will need ability to change the data value (a bool). However after the value changed, the filter on this property does not filter out changed item. For instance I have large dataset whose data has a property of Boolean named as 'Ignore'; I create a dimension for this property to filer out the data have ignore= true. In app, user has a capability to change this value from false to true. after value changed true, clear filters and refilter the data on this dimension, I am still getting the data item which supposed being filterred out. Is it by design or there is way to around it?
From the API Docs and source it appears this behaviour is not supported.
https://github.com/square/crossfilter/wiki/API-Reference