Move items in a QSortFilterProxyModel

JonB

I believe your range is wrong, at least in ::moveDown. itemIndex + 2 doesn't seem right for a move by one row.... E.g. when itemIndex == rowCount() - 2 it will produce rowCount(), which is too high for beginMoveRows()?
beginMoveRows() can be passed more rows than you're actually moving. While debugging I would pass everything from 0 to rowCount() - 1 so that you can eliminate bad parameters to beginMoveRows() from being the issue.

You are quite sure that with your QSortFilterProxyModel you are not imposing any kind of sort? Because adding/moving rows when Qt has sorting on always causes problems.

Valerian

@JonB
Yes I'm sure I'm not used any kind of sort. The move up function seems to select the next item but doesn't actually physically move it.

JonB

@Valerian

That's why I suggested you start by setting the range to all rows and see how you go.
beginMoveRows() returns a bool to indicate whether it likes your parameters. Check this! And carefully read the discussion of moving up & down in http://doc.qt.io/qt-5/qabstractitemmodel.html#beginMoveRows.

Valerian

@JonB said in Move items in a QSortFilterProxyModel:

That's why I suggested you start by setting the range to all rows and see how you go.

I tried setting the following for the moveUp

if(itemIndex > 0 && itemIndex < rowCount())
{
	qDebug() << "SSSSSSSSS " << beginMoveRows(QModelIndex(), 0, rowCount() - 1, QModelIndex(),
					  itemIndex - 1);
		moveRow(QModelIndex(), itemIndex, QModelIndex(), itemIndex - 1);
		endMoveRows();
	}

and here's the response

SSSSSSSSS  false
10:14:16.649 18/07/2018 [FATAL] global\qglobal.cpp:3044 - ASSERT: "!this->isEmpty()" in file c:\users\qt\work\qt\qtbase\include\qtcore\../../src/corelib/tools/qstack.h, line 62
10:14:35.029 18/07/2018 [FATAL] global\qglobal.cpp:3044 - ASSERT: "asize >= 0 && asize <= aalloc" in file c:\users\qt\work\qt\qtbase\include\qtcore\../../src/corelib/tools/qvector.h, line 535
10:14:35.934 18/07/2018 [FATAL] global\qglobal.cpp:3044 - ASSERT: "uint(d->size) <= d->alloc" in file c:\users\qt\work\qt\qtbase\include\qtcore\../../src/corelib/tools/qvector.h, line 614
10:14:36.198 18/07/2018 [FATAL] global\qglobal.cpp:3044 - ASSERT: "asize >= 0 && asize <= aalloc" in file c:\users\qt\work\qt\qtbase\include\qtcore\../../src/corelib/tools/qvector.h, line 535
10:14:36.421 18/07/2018 [FATAL] global\qglobal.cpp:3044 - ASSERT: "uint(d->size) <= d->alloc" in file c:\users\qt\work\qt\qtbase\include\qtcore\../../src/corelib/tools/qvector.h, line 614
10:14:36.909 18/07/2018 [FATAL] global\qglobal.cpp:3044 - ASSERT: "!this->isEmpty()" in file c:\users\qt\work\qt\qtbase\include\qtcore\../../src/corelib/tools/qstack.h, line 62

JonB

@Valerian
Ignore my idea of passing "all" rows as range. Revert to whatever code you think is right for just one row. Then try again.

The important thing is my second point, the return result of beginMoveRows() with whatever parameters you give it. Your debug shows it returned false. So long as it returns false in your case, it's not going to work, and you need to figure why not!

Valerian

@JonB said in Move items in a QSortFilterProxyModel:

he important thing is my second point, the return result of beginMoveRows() with whatever parameters you give it. Your debug shows it returned false. So long as it returns false in your case, it's not going to work, and you need to figure why not!

I reverted the code and it returns true but the row isn't actually shifted

JonB

@Valerian
Fair enough. Then I can only suggest you read that doc page carefully to understand what they are saying is required in each direction, and play with the parameters/which row/how far you move the row till you see something happening, and figure from there.

VRonin

moveRow must already call beginMoveRows and endMoveRows inside itself so you are duplicating calls
moveRow is not implemented in any of Qt's model (as far as I'm aware) so it will probably just do nothing and return false (check the return value)

What model are you using as source to this proxy?

Valerian

@VRonin said in Move items in a QSortFilterProxyModel:

What model are you using as source to this proxy?

QAbstractTableModel

JonB

@VRonin
Don't forget OP is claiming that one of his moveUp/moveDown does work, while the other does not....

moveRow must already call beginMoveRows and endMoveRows inside itself so you are duplicating calls

If that's true that might explain....!

VRonin

@Valerian said in Move items in a QSortFilterProxyModel:

QAbstractTableModel

Did you reimplement moveRow?

Valerian

@VRonin I re-implemeted it in the QSortFilterProxyModel. Didn't want to move rows in the main model

VRonin

Can you post the implementation?

JonB

@Valerian
I now realise you got your code from https://stackoverflow.com/a/43683619/489865, didn't you? Wasn't actually confirmed to be correct solution. I still don't get where the + 2 comes from, but there you are.

IF @VRonin does not sort you out (so to speak :) ), I don't know, but maybe you should have a careful read of https://stackoverflow.com/a/12254935/489865. Not sure whether the following might affect you:

(2) Yes if you don't use sorting, just filtering. Qt in such case erases entire internal mapping and re-builds it from scratch so the data will be just as they appear in source model

Valerian

@VRonin said in Move items in a QSortFilterProxyModel:

Can you post the implementation?

I have already posted the implementation of the QSortFilterProxyModel. Are you seeking the implementation of the QAbstractTableModel?

VRonin

@VRonin said in Move items in a QSortFilterProxyModel:

Did you reimplement moveRow?

@Valerian said in Move items in a QSortFilterProxyModel:

I re-implemeted it in the QSortFilterProxyModel.

I don't see an implementation of FilterProxyModel::moveRow (or rather FilterProxyModel::moveRows) above

Valerian

@JonB said in Move items in a QSortFilterProxyModel:

I now realise you got your code from https://stackoverflow.com/a/43683619/489865, didn't you? Wasn't actually confirmed to be correct solution. I still don't get where the + 2 comes from, but there you are.

Yes I did pick it from there.

Thanks for the other stackflow link. Will have a read on that.