The ChartLegendElement displays the legend items in a StackLayoutElement which can be either vertical or horizontal.

        public RadForm1()
        {
            InitializeComponent();

            Random rand = new Random();
            List<LineSeries> list = new List<LineSeries>();
            for (var index = 1; index <= 15; index++)
            {
                LineSeries ls = new LineSeries();
                ls.LegendTitle = "Series " + index;
                list.Add(ls);
            }

            for (int index = 1; index <= 100; index++)
            {
                foreach (LineSeries s in list)
                    s.DataPoints.Add(new CategoricalDataPoint(index, rand.Next(0, rand.Next(5, 20))));
            }
            this.radChartView1.Series.AddRange(list.ToArray());
            this.radChartView1.ShowLegend = true;
            this.radChartView1.ChartElement.LegendPosition = LegendPosition.Bottom;
            this.radChartView1.ChartElement.LegendElement.StackElement.Orientation = Orientation.Vertical;
        }

Horizontal:

Vertical:

It would be good to provide an option for wrapping the legend items:

Workaround: use an appropriate container for the legend items to wrap the legend item and use the space more efficiently

https://docs.telerik.com/devtools/winforms/knowledge-base/chartview-wrap-legend-items 

The method GridViewTemplate.SetError creates in most situations an GridViewDataErrorEventArgs object twice.

Current code:

GridViewDataErrorEventArgs args = new GridViewDataErrorEventArgs(exception, 0, 0, GridViewDataErrorContexts.Commit);
if (e != null)
{
    args = new GridViewDataErrorEventArgs(exception, e.ColumnIndex, e.RowIndex, GridViewDataErrorContexts.Commit);
}

In assume in most cases e will not be null, so in must cases the first args will be removed. This has a small negative impact on memory and performace.

Suggestion:

GridViewDataErrorEventArgs args = e == null 
    ? new GridViewDataErrorEventArgs(exception, 0, 0, GridViewDataErrorContexts.Commit)
    : new GridViewDataErrorEventArgs(exception, e.ColumnIndex, e.RowIndex, GridViewDataErrorContexts.Commit);

This is about the following method:

        public void SetError(GridViewCellCancelEventArgs e, Exception exception)
        {
            GridViewDataErrorEventArgs args = new GridViewDataErrorEventArgs(exception, 0, 0, GridViewDataErrorContexts.Commit);
            if (e != null)
            {
                args = new GridViewDataErrorEventArgs(exception, e.ColumnIndex, e.RowIndex, GridViewDataErrorContexts.Commit);
            }
            this.EventDispatcher.RaiseEvent<GridViewDataErrorEventArgs>(EventDispatcher.DataError, this, args);

            if (args.ThrowException)
            {
                throw args.Exception;
            }

            if (args.Cancel)
            {
                //TODO: cancel row edit 
            }
        }

Right now, the method GridViewTemplate.SetError accepts a parameter of type GridViewCellCancelEventArgs, which in itself is weird, because event args should only be used inside events and OnXXX-methods. But since SetError fires an event, one could argue that this method is like a OnXXX-method.

But inside it becomes more weird, it translates GridViewCellCancelEventArgs into GridViewDataErrorEventArgs. And I must admit, the last DataError-args feel a lot more logical that CellCancel-args when firing setting an error. Furthermore, if I create an override of class GridViewCellCancelEventArgs to contain more data about an error, this information never reaches the event handlers.

So why this translation? Or better: Why this parameter?

My suggestion would be to make a new overload:

public void SetError(GridViewDataErrorEventArgs e)
{
    if (e == null)
        throw new ArgumentNullException(nameof(e));

        this.EventDispatcher.RaiseEvent<GridViewDataErrorEventArgs>(EventDispatcher.DataError, this, args);

        if (args.ThrowException)
                throw args.Exception;

        if (args.Cancel)
        {
                // TODO: I really do not know what telerik wanted to do here, so I live this up to Telerik.
        }
}

Repro-steps

1. Create a Form
2. Add a RadGridView control
3. Add two Text columns
4. Run the form
5. Add a new row to the RadGridView with values "a" and "b"
6. Add a another row to the RadGridView with values "c" and "d"
7. Select the first row and copy the row values
8. Select the first cell of the second row and paste the values

Expected behavior

• The values "a" and "b" are in the second row

Observed behavior

• The values "ab" and "d" are in the second row

Repro-steps

1. Create a Form
2. Add a RadGridView control
3. Add two Text columns
4. Run the form
5. Add a new row to the RadGridView with values "a" and "b"
6. In Excel fill a row with two cells with values "c" and "d"
7. Select those cells and copy them to the clipboard
8. In the RadGridView select the first cell of the row
9. Paste the data

Expected behavior

• The values "c" and "d" are in the row

Observed behavior

• The values "System.IO.MemoryStream" and "b" are in the row

Extra info

If the first column is of another type (like GridViewDecimalColumn) a DataError occurs (because "System.IO.MemoryStream" cannot be parces to a decimal).

To replicate the missing button when the application is run on my main monitor with 150% DPI scaling:

If the RadControl.EnableRadAutoScale property is set to false in the Program.cs file, the button is placed as expected:



When I replace the form to inherit from RadForm, not the MS Form, the button is clipped:

Searching for a workaround for a previously reported bug, in walked into a little thing:

private void SelectAllCells()
    {
      GridViewRowInfoEnumerator rowInfoEnumerator = new GridViewRowInfoEnumerator((IHierarchicalRow) this.GridViewElement.Template);
      List<GridViewCellInfo> gridViewCellInfoList = new List<GridViewCellInfo>();
      this.MasterTemplate.SelectedCells.BeginUpdate();
      this.MasterTemplate.SelectedCells.Clear();
      while (rowInfoEnumerator.MoveNext())
      {
        GridViewRowInfo current = rowInfoEnumerator.Current;
        if (current.CanBeSelected)
        {
          foreach (GridViewCellInfo cell in current.Cells)
            this.MasterTemplate.SelectedCells.Add(cell);
        }
      }
      this.MasterTemplate.SelectedCells.EndUpdate(true);
    }

I am rewriting the RadGridViewPaste-mechanism. Therefor I want to call PastingCellClipboardContent-event using the EventDispatcher.CellClipboardPaste. therefor I need the class GridViewCellValueEventArgs. It has two constructors:

1. A constructor with indexes. If this one is used, properties RowInfo and Column will always be null.
2. A constructor which accepts a row and column as parameters. All properties are set and usable.

The second constructor is marked as Internal, not Public. 

Request: Can this constructor be made Public?

1. Set ThemeResolutionService.AllowAnimations = false;

2. Run the attached project and set a filter via the GridFilterButton popup

3. Click "Clear filter" option in the GridFilterButton popup

You will see that the GirdFilterButton visual indication is of an applied filter. However, there is no applied filter.

Implement functionality to rotate the text in a cell.

I have made an override for class GridViewDataRowInfo (called GridViewDataRowInfoPlus)

Every time a new row is created by RadGridView, I expect OnCreateRowInfo to be called, so I can return my own instance of GridViewDataRowInfoPlus.

This goes well until a user enters new data in a new row and when this row is added the grid, OnCreateRowInfo is not being called and an instance of GridViewDataRowInfo is added to the grid.

For me that is unexpected and undesirable. Maybe I am doing it wrong. In that case: What is the correct way to intercept the row-creation-process and return instances of my own type?

The version of Telerik products is requested as information where submitting a support ticket. In case one forgot the version currently installed, there is no way to find it (may be I am the only one who don't know). Usually most software provide such information in the "About" box. I found it difficult and not very useful that the "About" menu of Telerik connect to the Progress Home Page instead of displaying some useful information about the current installation (user name, version, installed components, license status, etc....). There are already many other actions (request support, online documentation, etc...) that connect to the web site. The "About" box should just be displayed within VS and provide information on the installation. Or at least you link it to the Progress Control Panel and adding a page for information I mentioned and more that user may need to know when needed.

Below is an example from VS.

This bug is not about RadGridView, but about this ticketing system. I simply did not know where to send it.

Repro steps

1. Go to a ticket with a question/answer with a link inside of it (for example, ticket: https://feedback.telerik.com/winforms/1573736-radgridview-oncreaterowinfo-not-being-called-when-user-is-adding-a-new-row) 
2. Write a nice response with bold texts, paragraphs, etc. Do not send it yet.
3. In the original text, press on the link. You leave the page.
4. Right now be very scared that you might have lost your answer and press BACK.
5. You go back to the back where you were writing your answer.

Expected behavior

• Your answer is still there as it was, including all formatting.

Observed behavior

• Your answer is still there, but all formatting has been replaced by HTML-tags, and no button or way to convert this HTML to formatted text.

Repro-steps

1. Create a RadGridView
2. Fill it with 8000 rows
3. Select all rows
4. Delete all rows and measure the time
5. Repopulate the same grid with the same 8000 rows
6. Sort on a column
7. Select all rows
8. Delete all rows and measure the time

Expected behavior

• Both measured times are the same

Observed behavior

• Deleting rows when a column is sorted take muuuuuuuuch more time

The problem is, after each (!!!) deleted row (not after all deleted rows) the sort-routine kicks in. Sorting all remaining rows. That is 7999 times to many.

One can argue that sorting after deleting something is not required at alle, since the order of the remaining rows (in this case none, but in one cases maybe more) will never change.

I also noticed a HybridIndex is used, possibly to increase performance during adding. Somehow it might hinder performance during deleting stuff.