Changes since Alpha 1

Class Reference

Latest update.

Packages

Changed dynamic enabling and disabling of DEOs so it is a little less aggressive. The new procedure SetDynamicEnabled is called during each DEO's refresh to dynamically set a DEO's enabled_state. This was added to added to support DD_COMMIT but it was getting in the way of custom settings that developers were applying to their DEOs. Now the enabling is only done if the DEO is committable (i.e., if its DD field's DD_COMMIT is set or it is a no-switch if committed parent). Since no existing DDs support these new features, no existing applications should change.

Studio

Changed combo's window style in WSDL parser so you can use URLs that are longer than the width of the combo (from Vincent).

Packages

1. Fixed problem with dbRadioGroups raising an error when Clear_all is selected (side effect of DD_Commit support).
2. Fixed problem where dbRadiogroups were not properly enabling/disabling themselves based on the current field commit status.
3. Fixed problem in cDbCJGrid where using a text field would raise an error (side effect of extended field support)
4. Fixed problem where a change to an text field may not get saved the first time because the extended DD field is getting creating and clearing the field changed state.
5. Fixed problem with Null Parent Support where various validation errors would get fired on a Null parent.
6. Fixed problem where changing a key field value and then clearing the form would result in an error upon navigation (side effect of Null Parent Support)
7. Changed a couple of DataDictionary meta tags as per discussion with Dennis.

Examples

Fixed anchors for XML sample views.

dfctrls17.0 Build 2

This was inadvertently left out of Alpha 1.
This changes the way the Windows date mask handles years - specifically what happens when the year is omitted. For example, if the regional mask was dd/mm/yyyy or mm/dd/yyyy and the input operator entered "1/1/". Prior to this change a missing year was treated as 0 and then the epoch year logic was applied, turning the date into 1/1/2000. This would do the same as entering "1/1/0". In 2011, entering "1/1/" and getting "1/1/2000" is never what you want, so an input operator would always have to type in the year.

This is now changed so that a missing year digit is interpreted to mean current year, so that date would currently be treated as 1/1/2011. It would be the same as typing "1/1/11" or "1/1/2011". What this means is that data entry operators could omit the year and they probably get the default year they expect.

This does change existing behaviors in application but this seems like a good change. There is one other change that might be different. If someone is not using the Epoch date conversion logic and actually still using 2 digit dates in their embedded database, leaving off the year will result in a four year date (it still will be 2011). I can't imagine that any is using current VDF versions and still using 2 digit dates eleven years after Y2K.

When masked dates are entered, the input mask is always the windows regional setting; it doesn't matter what your actual display mask is. I expect this to work properly with US (mm/dd/yyyy) and European formats (dd/mm/yyyy). I think this might also work with military formats (yyyy/mm/dd) if you just skipped the year (e.g., started by pressing "/") although I don't know how useful this is. I also don't know if there is a military regional setting.