Is Qt’s model–view system intended to work with one model or multiple models?

Question

I’m trying to display some tabular data with a QTableView subclass and a QAbstractTableModel subclass. I can’t get the data to show up, but before I start really pounding on it I want to make sure that I’m using models in the way they were intended.

The data layer of my application periodically receives new data and distributes the data to the other parts of the application by calling slots like

void new_data_received(QSharedPointer<Measurement> measurement)

where Measurement is my data class. This allows the data to be passed around without being copied (some of my data classes are very large). Measurements are immutable; the table view that displays them doesn’t allow any editing.

Measurement is a subclass of QAbstractTableModel, so whenever I receive a new measurement I call set_model on my QTableView subclass instance with the new data as a parameter. (In the time before the first measurement is received there is no model set on the table view.)

Are Qt’s view classes intended to be used like this, with a new model being set every so often? Or should there be just one instance of the model class, with the same lifetime as the table view, that receives the new data and emits dataChanged? The latter seems like it adds unnecessary structure—at least in my case—but maybe that’s the way the system was designed to be used.

Answer 1

I don't think your Measurement class should be a subclass of QAbstractTableModel. It should represent raw data instead. So maybe a struct with some parameters or a list of structs will be a right type for your data class. Then you should implement a custom model where incoming data are added to. So, when new data arrives that model will automatically update all the views connected to it. In this case new data affects directly your model only, not the views.

I suppose resetting view's model every time is not the right way to do what you want.