Best practice for controls on updated properties @ L.V. Expertise X3

Best practice for controls on updated properties

The best practices for controls can be summarized in the following bullets:
* If the validity of a property is not linked to the value of another property that can be modified in the same transaction, the control can be done on the control rule of the property.
* If a validity of a property depends on the value of other properties, the control must be done globally at the class level.

The reason for this rule is that you cannot assess in which order the properties have been entered and controlled.

If you want to perform such controls immediately after data entry to enhance the user experience, you can add such code in the UI layer (attached to the representation); however:

The entry order might not be the one you expect.
You have to perform the global control because the corresponding code can be invoked in service mode. The validity of the rules must also be ensured even in this mode.

For example, if you want to control that no duplicates values exist on a row in a grid, you can write it in the global control code for the main class.

If uni(this.COLLECTION(1..maxtab(this.COLLECTION)).MYKEY)<>0# A duplicate key exists for the rowARET= Fmet this.ASETERROR .... # Place here the right error messageEnd ARETEndif

At first view, this code might work. Using an aggregation operator (such as uni, min, max...) on a collection is the correct thing to do. Unfortunately, there is a flaw in this code because the deletion of lines in a collection is managed logically by setting a value equal to [V]CST_ADEL or [V]CST_ANEWDEL in the ASTALIN property of the collection. This means that trying to insert a new key that corresponds to a previously deleted value will also display the error.

The example must be modified as follows:

Local Char KEY(20)(1..)Local Integer LINES, CURLIN, L_INDEX, I, J# Let's make a list of the existing codesFor CURLIN=1 to maxtab(this.COLLECTION)If this.COLLECTION(CURLIN)<>Null and find(this.COLLECTION(CURLIN).ASTALIN,[V]CST_ADEL,[V]CST_ANEWDEL)=0LINES+=1KEY(LINES)=this.COLLECTION(CURLIN).MYKEYL_INDEX(LINES)=CURLINEndifNext CURLIN# Now we can test the unicity of the codesJ=uni(KEY(1..LINES))If J<>0# A duplicate key exists for the rowI=find(KEY(J),KEY(1..J-1))# L_INDEX(I) is the first duplicate key position, and L_INDEX(J) the second positionARET= Fmet this.ASETERROR .... # Place here the right error messageEnd ARETEndif

If you write this code in the script associated with the collection, you would write it as follows:

Local Char KEY(20)(1..)Local Integer LINES, CURLIN# Let's make a list of the existing codesFor CURLIN=1 to maxtab(this.APARENT.COLLECTION)If this.APARENT.COLLECTION(CURLIN)<>Null and find(this.APARENT.COLLECTION(CURLIN).ASTALIN,[V]CST_ADEL,[V]CST_ANEWDEL)=0LINES+=1KEY(LINES)=this.APARENT.COLLECTION(CURLIN).MYKEYEndifNext CURLINIf uni(KEY(1..LINES))<>0# A duplicate key exists for the rowARET= Fmet this.ASETERROR .... # Place here the right error messageEnd ARETEndif

Also, this would be risky from performance point of view because it would be repeated on every line in service mode.

Index Home