[SalesForce] Recursive Trigger / Trigger Context Issue

So there are tons of questions on here an d on the boards around using static variables to keep trigger from recursively calling itself. There also is a lot of information out there around best practices in triggers. These bets practices include only having one trigger per object and executing all logic outside of the trigger itself in a helper class. My issue seems to be trying to merge these two best practices. Helper classes for triggers, and use of static variables to prevent recursive trigger calls

Just for simplicity sake, lets assume the logic in the helper classes is ok. It's bulkified, its optimized, etc. Here is my Account Trigger

trigger AccountTrigger on Account (before insert, after insert, before update, after update) {

    AccountTriggerHelper helper = new AccountTriggerHelper();

    if(trigger.isBefore && trigger.isUpdate){
        helper.updateAccountUpdateSummary(trigger.new, trigger.old, trigger.newMap, trigger.oldMap);    
    }   

    if(trigger.isAfter){
        if(trigger.isUpdate){
            helper.maintainCAAccountTeamMembers(trigger.new, trigger.old, trigger.newMap, trigger.oldMap, trigger.isInsert, trigger.isUpdate);
            helper.updateCortextOpportunityOwners(trigger.new, trigger.old, trigger.newMap, trigger.oldMap);
        }
        if(trigger.isInsert){
            helper.maintainCAAccountTeamMembers(trigger.new, trigger.old, trigger.newMap, trigger.oldMap, trigger.isInsert, trigger.isUpdate);
        }
    }
}

As you can see I just have a helper class that I call the methods to execute the logic for the trigger. These all work great if I have them separated (each their own trigger). v The issue is I have test code that when I run is erroring out with a 'Too many SOQL queries' error.

When I check the debug logs, it seems that my triggers are being called recuresively, and it keeps going until it has run enough times in the same context to get the too many SOQL queries error.

So I thought I would use my TriggerContextUtility class to help with this. Its just a class with a static variable to keep track of the first run of the trigger. This works fine in my triggers with only one method being called.

Heres the triggerContetUtility class

public class TriggerContextUtility {

    private static boolean firstRun = true;

    public static boolean isFirstRun() {
        return firstRun;
    }
    public static void setFirstRunFalse(){
        firstRun = false;
    }
}

Very simple. Normally, I would use this utility class like below to ensure that the trigger only fires once per context, but adding it like below doesn't work

trigger AccountTrigger on Account (before insert, after insert, before update, after update) {

    AccountTriggerHelper helper = new AccountTriggerHelper();
    if(TriggerContextUtility.isFirstRun()){
        if(trigger.isBefore && trigger.isUpdate){
            helper.updateAccountUpdateSummary(trigger.new, trigger.old, trigger.newMap, trigger.oldMap);    
        }   

        if(trigger.isAfter){
            if(trigger.isUpdate){
                helper.maintainCAAccountTeamMembers(trigger.new, trigger.old, trigger.newMap, trigger.oldMap, trigger.isInsert, trigger.isUpdate);
                helper.updateCortextOpportunityOwners(trigger.new, trigger.old, trigger.newMap, trigger.oldMap);
            }
            if(trigger.isInsert){
                helper.maintainCAAccountTeamMembers(trigger.new, trigger.old, trigger.newMap, trigger.oldMap, trigger.isInsert, trigger.isUpdate);
            }
        }
    }
    TriggerContextUtility.setFirstRunFalse();
}

This does not work however. It seems my before methods are called, but not my after methods. I have tried moving the TriggerContextUtility.setFirstRunFalse(); into different places, but nothing seems to correctly fire all methods.

So I either leave out my static variable check and the trigger is recursively calling itself until it errors out. OR, I fix this recursive call using my TriggerContextUtility class, but then not all methods within the trigger are being called.

Can anyone offer any insight as to what I am doing wrong here? Am I missing something. This seems like something that should be simple, but for some reason I can't get it to work. Hoping for another set of eyes to find the flaw in my logic here.

EDIT

Here is the debug statements I added to the trigger

trigger AccountUpdatesTrigger on Account (before insert, after insert, before update, after update) {

    AccountTriggerHelper helper = new AccountTriggerHelper();
    if(TriggerContextUtility.isFirstRun()){
        if(trigger.isBefore && trigger.isUpdate){
            helper.updateAccountUpdateSummary(trigger.new, trigger.old, trigger.newMap, trigger.oldMap);    
        }   
        *system.debug('First Run Value Line 8: ' + TriggerContextUtility.isFirstRun());*

        if(trigger.isAfter){
            if(trigger.isUpdate){
                *system.debug('First Run Value Line 11: ' + TriggerContextUtility.isFirstRun());*
                helper.maintainCAAccountTeamMembers(trigger.new, trigger.old, trigger.newMap, trigger.oldMap, trigger.isInsert, trigger.isUpdate);
                helper.updateCortextOpportunityOwners(trigger.new, trigger.old, trigger.newMap, trigger.oldMap);
                *system.debug('First Run Value Line 14: ' + TriggerContextUtility.isFirstRun());*
            }
            if(trigger.isInsert){
                *system.debug('First Run Value Line 17: ' + TriggerContextUtility.isFirstRun());*

                helper.maintainCAAccountTeamMembers(trigger.new, trigger.old, trigger.newMap, trigger.oldMap, trigger.isInsert, trigger.isUpdate);
            }
            TriggerContextUtility.setFirstRunFalse();
        }
    }   
}

The line 14 debug statement is outputting FALSE. Why would that be?

Best Answer

I don't think you want the ContentUtility check to be at the Trigger level - I typically refer to it in my helper class, or even at the individual method level in that class. There may well be instances where you want one method to run twice if something was updated, but not others. So I would move the check for TriggerContextUtility into the helper class and I think it will work fine...

As an update - I have always used these posts as my guide.

http://gokubi.com/archives/two-interesting-ways-to-architect-apex-triggers http://www.embracingthecloud.com/2010/07/08/ASimpleTriggerTemplateForSalesforce.aspx

There are (many!) more variations, but I would keep simple at this stage... – BritishBoyinDC 1 min ago edit

Related Solutions

[SalesForce] Update same record in after insert/update in a trigger context

In a Before Trigger, Trigger.new and Trigger.old both contain the same values. However, in a Before Trigger, one can operate on the values in Trigger.new and change their values.

In an After Trigger, Trigger.new will contain the the Id's and field values of any records that have been changed by a DML operation. The values contained in Trigger.new can be compared to Trigger.old in order to determine which records meet the criteria for firing the trigger and need to be operated on. Unlike in a Before Trigger, the values in Trigger.new cannot be changed or edited.

Recursion is frequently handled by setting a boolean static variable to "true" when a trigger initially fires. Upon entry to the trigger, that static variable is tested. If the boolean is "true" when re-entered in the same execution context, then the trigger is exited rather than allowing it to run again. There are a variety of methods for handling recursion and this is probably the most common.

Edit

I think what you're looking for are trigger patterns such the one described by Dan Appleman in his book Advanced Apex Programming for Salesforce and Force.com (Note: sample code available from App Exchange as an unmanaged package), Hari Krishan's Architecture Framework to Handle Triggers or Kevin O'Hara's Trigger Framework among others that exist or have been espoused which you can find through searching that go beyond the Trigger Pattern for Tidy, Streamlined, Bulkified Triggers.

Using a Main Dispatcher class, these approaches take charge of trigger execution, extending it to an enterprise level application that's scalable, can have it's own diagnostics classes and error reporting, utilize common helper classes and much more. One can also control trigger execution order; something that can't be done with triggers that operate independently of one other without some kind of central control for each object or, as is the case with some architures, potentially for related objects as well.

Discussing them in depth is beyond the scope of this forum, but I'll point you to at least a few posts that you may find relevant including: Controlling Trigger Execution, One Trigger Approach: Any Logic in Trigger?, Difficulties with Appleman's Trigger Classes and General trigger bulkification - best practices.

[SalesForce] Why Trigger.newMap in before Update event

New map, as the name suggests, is to provide access to the new values the records will take on during the DML operation.

You might want this in before events to perform some kind of validation, for example (pseudeocode):

for each record in trigger.old
  if record.field == 'A' && trigger.newMap.get(record.Id).field == 'C'
    error('You can not change field straight from A to C!');

Using newMap in a Before Update trigger will let you access the new field values, oldMap will provide you with the existing values. As you can see from the above, it's useful to have access to both so that you can take action as business logic dictates.

Best Answer

Related Solutions

[SalesForce] Update same record in after insert/update in a trigger context

[SalesForce] Why Trigger.newMap in before Update event

Related Topic