Here is a minimal test case. I set up my ultragrid with an ultradatasource with two decimal columns and two rows with 0s. I suggest having one cell not have a 0, otherwise there is no way to clear the cache. (I am using version 12.1.20121.2008)

   public partial class Form1 : Form {
        public Form2() {
            InitializeComponent();
            DF df = new DF();
            foreach(UltraGridColumn uc in ultraGrid1.DisplayLayout.Bands[0].Columns) {
                uc.Editor.DataFilter = df;
            }
        }
    }
    class DF : IEditorDataFilter {
        public object Convert(EditorDataFilterConvertArgs conversionArgs) {
            switch(conversionArgs.Direction) {
                case ConversionDirection.EditorToDisplay:
                    UltraGridCell cell = (UltraGridCell)conversionArgs.Context;
                    if(cell.IsActiveCell && conversionArgs.Value is decimal) {
                        conversionArgs.Handled = true;
                        return ((decimal)conversionArgs.Value).ToString("0.0");
                    }
                    break;
            }
            return conversionArgs.Value;
        }
    }

Thank you (belatedly ^^;) for taking the time to respond.

I cannot find a way to differentiate between a cell being in edit mode or not with this method because it seems the first time I enter my filter both conversionArgs.Editor.IsInEditMode and ((UltraGridCell)conversionArgs.Context).IsInEditMode return false, and I don't seem to enter the filter when one is true (likely because it is using the cache at that point). Therefore, it seems I still need to clear the cache at least once, unless there is some other way you suggest my being able to differentiate when the activated cell is or is not going into edit mode.

Hi Jon,

Any cell that is activated will go into edit mode, unless is cannot.

There are several reasons why a cell would not be able to enter edit mode.

1) You set a property on that cell that disallows this. So you could check the various Activation/CellActivation properties on the row, cell, and column.

2) The data source does not allow editing.

3) The BeforeEnterEditMode event is cancelled.

You can detect the first two, but not the third one - since you can never be sure if your event handler fires before or after any other event listener.

It seems like the conversionArgs.Editor.IsInEditMode should work, though. Wasn't that working before? Or were you simply checking IsActive on the Cell before?

Mike,

The cell does go into edit mode. However at the time it enters my DataFilter, it seems is not yet in edit mode. Then, once the editor is in edit mode, the cache takes over and my DataFilter is no longer used (until the editor leaves edit mode).

I had been testing if the cell was active and the grid was focused before for reasons I cannot recall ^^. At that time, I had not checked what would happen if the cell was active but not in edit mode (e.g. via pressing F2).

Hm. Maybe assigning the Editor to the cell is triggering the DataFilter. So it's getting hit in BeforeCellActivate, before the cell has entered edit mode. I guess you could try moving the BeforeCellActivate code to AfterCellActivate or BeforeEnterEditMode. But I think I don't think this will work, since they both fire before the cell is in edit mode.

Does it really matter if the cell is in edit mode? Maybe you can just apply your filter to the active cell. In that case, you don't even need to the DataFilter to do ANY checking, since the DataFilter is only applied to the active cell, anyway.

Mike Saltzman"]

Does it really matter if the cell is in edit mode?

Perhaps not. It may just be a personal preference. If I'm about to edit a value, then I don't mind so much if it is displayed a bit differently when I click on it, but if I'm just moving around the grid with the keyboard instead of a mouse, then it doesn't seem as nice an interface to me.

As far as the caching, since it is only in effect if the editor is in edit mode, then I imagine I already lose a lot of the benefit if I'm always assigning a new editor to the cell being edited, anyway.

Thank you for your time on this subject. I'm experimenting with a hybrid of the BeforeCellActivate/Deactivate and the EditorWithText child class that clears the EditorToDisplay cache in calls to GetElementText. I imagine it would work the same without the events, but I feel more comfortable isolating the editor to the single cell since I'm using undocumented features.