QTableWidget segmentation fault with setTextAlignment

[Publicnamer]

I really hope someone is fixing QTableWidget bugs, because this simply shouldn't happen IMHO:

            QModelIndex index = table->model()->index(row, 0);
            table->model()->setData (index, "?", Qt::DisplayRole);
            QTableWidgetItem *item = table->item (row, 0);
            item->setTextAlignment (Qt::AlignRight);
                             ^-- crash here

The table has 100 rows and 100 columns.

[JoeCFD]

do you check if item is nullptr?

[VRonin]

@Publicnamer said in QTableWidget segmentation fault with setTextAlignment:

^-- crash here

It should not. What version of Qt are you using? What does table->model()->setData return?

[JonB]

@Publicnamer
As @VRonin has asked you.

I would assume QModelIndex index = table->model()->index(row, 0); returns an invalid index if out of range, but QTableWidgetItem *item = table->item (row, 0); would have to return a nullptr....

Alternatively, you may have a valid index(row, 0). But that does not mean there is a QTableWidgetItem *item at table->item (row, 0), cells start out with no QTableWidgetItem in them, you have to put one in, IIRC.

Which would explain behaviour you see without there being any " fixing QTableWidget bugs" ?

[VRonin]

@JonB said in QTableWidget segmentation fault with setTextAlignment:

you have to put one in

setData puts the item in so it should be available afterwards

[JonB]

@VRonin
[Totally out of my head, not tested.] My understanding was that if you use the setData() level, that's fine, but I thought that did not create a QTableWidgetItem item for it. QTableWidget does not reimplement setData(), so you're saying it would have to do it on dataChanged() signal? (And that doesn't get emitted if you're not actually changing the value.)

EDIT OK looked on woboq, agreed, it has internal QTableModel::setData() which does as you say about creating the item.