MatterAI - Frontier LLM Intelligence at 70% Less Cost

Apex Anti-Patterns Overview

Apex, despite being a powerful language for Salesforce development, has several common anti-patterns that can lead to performance issues, governor limit violations, maintainability problems, and bugs. Here are the most important anti-patterns to avoid when writing Apex code.

SOQL Queries Inside Loops

// Anti-pattern: SOQL query inside a loop
for (Account acc : [SELECT Id, Name FROM Account LIMIT 100]) {
    // For each account, find its contacts
    List<Contact> contacts = [SELECT Id, FirstName, LastName 
                             FROM Contact 
                             WHERE AccountId = :acc.Id];
    
    // Process contacts
    for (Contact con : contacts) {
        // Do something with each contact
    }
}

// Better approach: Use relationship queries or maps
// Option 1: Use relationship query
for (Account acc : [SELECT Id, Name, 
                    (SELECT Id, FirstName, LastName FROM Contacts)
                    FROM Account LIMIT 100]) {
    // Process contacts
    for (Contact con : acc.Contacts) {
        // Do something with each contact
    }
}

// Option 2: Use a single query with a map
Map<Id, Account> accountMap = new Map<Id, Account>([SELECT Id, Name FROM Account LIMIT 100]);
List<Contact> allContacts = [SELECT Id, FirstName, LastName, AccountId 
                           FROM Contact 
                           WHERE AccountId IN :accountMap.keySet()];

// Group contacts by AccountId
Map<Id, List<Contact>> contactsByAccount = new Map<Id, List<Contact>>();
for (Contact con : allContacts) {
    if (!contactsByAccount.containsKey(con.AccountId)) {
        contactsByAccount.put(con.AccountId, new List<Contact>());
    }
    contactsByAccount.get(con.AccountId).add(con);
}

// Process accounts and their contacts
for (Account acc : accountMap.values()) {
    List<Contact> contacts = contactsByAccount.get(acc.Id);
    if (contacts != null) {
        for (Contact con : contacts) {
            // Do something with each contact
        }
    }
}

Never put SOQL queries inside loops. This can quickly exceed the governor limit of 100 SOQL queries per transaction. Instead, use relationship queries to retrieve related records in a single query, or use a map-based approach to fetch all related records in one query and then process them in memory.

DML Operations Inside Loops

// Anti-pattern: DML inside a loop
for (Account acc : [SELECT Id, Name FROM Account WHERE Rating = 'Hot']) {
    acc.Description = 'Hot account - high priority';
    update acc; // DML inside loop
}

// Better approach: Collect records and perform bulk DML
List<Account> accountsToUpdate = new List<Account>();

for (Account acc : [SELECT Id, Name FROM Account WHERE Rating = 'Hot']) {
    acc.Description = 'Hot account - high priority';
    accountsToUpdate.add(acc);
}

// Single DML operation outside the loop
if (!accountsToUpdate.isEmpty()) {
    update accountsToUpdate;
}

Avoid performing DML operations (insert, update, delete, upsert) inside loops. This can quickly exceed the governor limit of 150 DML operations per transaction. Instead, collect all records that need to be modified and perform a single bulk DML operation outside the loop.

Not Bulkifying Triggers

// Anti-pattern: Non-bulkified trigger
trigger ContactTrigger on Contact (after insert) {
    // Process one record at a time
    for (Contact con : Trigger.new) {
        // Query account for each contact
        Account acc = [SELECT Id, Name FROM Account WHERE Id = :con.AccountId];
        
        // Create a task for each contact
        Task t = new Task();
        t.Subject = 'Follow up with new contact';
        t.WhatId = acc.Id;
        t.WhoId = con.Id;
        insert t; // DML inside loop
    }
}

// Better approach: Bulkified trigger
trigger ContactTrigger on Contact (after insert) {
    // Collect all Account IDs
    Set<Id> accountIds = new Set<Id>();
    for (Contact con : Trigger.new) {
        if (con.AccountId != null) {
            accountIds.add(con.AccountId);
        }
    }
    
    // Query accounts in bulk
    Map<Id, Account> accountMap = new Map<Id, Account>(
        [SELECT Id, Name FROM Account WHERE Id IN :accountIds]
    );
    
    // Create tasks in bulk
    List<Task> tasksToInsert = new List<Task>();
    for (Contact con : Trigger.new) {
        if (con.AccountId != null && accountMap.containsKey(con.AccountId)) {
            Task t = new Task();
            t.Subject = 'Follow up with new contact';
            t.WhatId = con.AccountId;
            t.WhoId = con.Id;
            tasksToInsert.add(t);
        }
    }
    
    // Insert tasks in bulk
    if (!tasksToInsert.isEmpty()) {
        insert tasksToInsert;
    }
}

Always bulkify your triggers to handle multiple records efficiently. Salesforce can process up to 200 records in a single transaction, and your triggers should be designed to handle this scenario without exceeding governor limits. Use bulk queries and DML operations, and avoid processing records one at a time.

Hardcoding Record IDs

// Anti-pattern: Hardcoding record IDs
public void assignToDefaultQueue() {
    // Hardcoded ID of a specific queue
    Id queueId = '00G1a000001XgXx';
    
    List<Case> cases = [SELECT Id FROM Case WHERE Status = 'New'];
    for (Case c : cases) {
        c.OwnerId = queueId;
    }
    update cases;
}

// Better approach: Query for IDs or use custom metadata/settings
public void assignToDefaultQueue() {
    // Query for the queue ID
    Group defaultQueue = [SELECT Id FROM Group 
                         WHERE Type = 'Queue' AND DeveloperName = 'DefaultCaseQueue' 
                         LIMIT 1];
    
    List<Case> cases = [SELECT Id FROM Case WHERE Status = 'New'];
    for (Case c : cases) {
        c.OwnerId = defaultQueue.Id;
    }
    update cases;
}

// Even better: Use custom metadata or custom settings
public void assignToDefaultQueue() {
    // Retrieve queue ID from custom setting
    Queue_Settings__c settings = Queue_Settings__c.getInstance('Default');
    Id queueId = settings.Default_Case_Queue_ID__c;
    
    List<Case> cases = [SELECT Id FROM Case WHERE Status = 'New'];
    for (Case c : cases) {
        c.OwnerId = queueId;
    }
    update cases;
}

Never hardcode record IDs in your code. IDs can be different across environments (sandbox, production), and hardcoded IDs make your code environment-specific and brittle. Instead, query for the IDs dynamically or store them in custom metadata types or custom settings.

Not Using the Limits Class

// Anti-pattern: Not checking limits
public void processAccounts() {
    List<Account> accounts = [SELECT Id FROM Account LIMIT 50000];
    // Process accounts without checking limits
}

// Better approach: Check limits and handle accordingly
public void processAccounts() {
    // Check if we have enough SOQL queries available
    if (Limits.getQueries() >= Limits.getLimitQueries() - 1) {
        throw new CustomException('Too many SOQL queries used in this transaction');
    }
    
    // Check if we can process all accounts or need to batch
    Integer recordLimit = Limits.getLimitHeapSize() / 10000; // Approximate size per record
    List<Account> accounts = [SELECT Id FROM Account LIMIT :recordLimit];
    
    // Process accounts
}

Use the Limits class to check governor limits at runtime and handle situations where you might exceed them. This is especially important for code that processes variable amounts of data or is called from multiple contexts.

Not Using Try-Catch Blocks

// Anti-pattern: No exception handling
public void createAccount(String name) {
    Account acc = new Account(Name = name);
    insert acc; // Will throw exception if insert fails
    
    // Create related records
    Contact con = new Contact(LastName = 'Contact for ' + name, AccountId = acc.Id);
    insert con; // Might not be reached if account insert fails
}

// Better approach: Use try-catch for proper error handling
public void createAccount(String name) {
    try {
        Account acc = new Account(Name = name);
        insert acc;
        
        // Create related records
        Contact con = new Contact(LastName = 'Contact for ' + name, AccountId = acc.Id);
        insert con;
    } catch (DmlException e) {
        // Log the error
        System.debug('Error creating account: ' + e.getMessage());
        
        // Rethrow as custom exception with more context
        throw new AccountCreationException('Failed to create account: ' + e.getMessage(), e);
    }
}

// Custom exception class
public class AccountCreationException extends Exception {}

Use try-catch blocks to handle exceptions gracefully. This allows you to provide better error messages, perform cleanup operations, and prevent partial transactions. Create custom exception classes for specific error scenarios to make error handling more structured.

Using System.debug for Production Logging

// Anti-pattern: Using System.debug for production logging
public void processOrders() {
    System.debug('Starting order processing');
    
    List<Order__c> orders = [SELECT Id, Status__c FROM Order__c WHERE Status__c = 'Pending'];
    System.debug('Found ' + orders.size() + ' orders to process');
    
    // Process orders
    for (Order__c ord : orders) {
        try {
            // Process order
            System.debug('Processed order: ' + ord.Id);
        } catch (Exception e) {
            System.debug('Error processing order: ' + e.getMessage());
        }
    }
}

// Better approach: Use a logging framework or Platform Events
public void processOrders() {
    Logger.info('Starting order processing');
    
    List<Order__c> orders = [SELECT Id, Status__c FROM Order__c WHERE Status__c = 'Pending'];
    Logger.info('Found ' + orders.size() + ' orders to process');
    
    // Process orders
    for (Order__c ord : orders) {
        try {
            // Process order
            Logger.info('Processed order: ' + ord.Id);
        } catch (Exception e) {
            Logger.error('Error processing order: ' + ord.Id, e);
            
            // Create error log record
            Error_Log__c log = new Error_Log__c(
                Object_Type__c = 'Order__c',
                Record_Id__c = ord.Id,
                Error_Message__c = e.getMessage(),
                Stack_Trace__c = e.getStackTraceString()
            );
            insert log;
        }
    }
}

Don’t rely on System.debug for production logging. Debug logs are not easily accessible in production and are only retained for a limited time. Instead, use a custom logging framework, custom objects to store logs, or Platform Events for more robust logging in production environments.

Not Using Test.startTest() and Test.stopTest()

// Anti-pattern: Not using Test.startTest() and Test.stopTest()
@isTest
private class OrderServiceTest {
    @isTest
    static void testProcessOrders() {
        // Create test data
        List<Order__c> testOrders = createTestOrders(100);
        insert testOrders;
        
        // Call the method to test
        OrderService service = new OrderService();
        service.processOrders();
        
        // Verify results
        List<Order__c> processedOrders = [SELECT Id, Status__c FROM Order__c WHERE Status__c = 'Processed'];
        System.assertEquals(100, processedOrders.size());
    }
}

// Better approach: Use Test.startTest() and Test.stopTest()
@isTest
private class OrderServiceTest {
    @isTest
    static void testProcessOrders() {
        // Create test data
        List<Order__c> testOrders = createTestOrders(100);
        insert testOrders;
        
        // Reset governor limits
        Test.startTest();
        
        // Call the method to test
        OrderService service = new OrderService();
        service.processOrders();
        
        // Force async processes to complete and restore governor limits
        Test.stopTest();
        
        // Verify results
        List<Order__c> processedOrders = [SELECT Id, Status__c FROM Order__c WHERE Status__c = 'Processed'];
        System.assertEquals(100, processedOrders.size());
    }
}

Always use Test.startTest() and Test.stopTest() in your test methods. These methods reset governor limits for the code executed between them and force any asynchronous processes to complete. This ensures that your tests accurately verify the behavior of your code under normal governor limits.

Not Using @TestVisible

// Anti-pattern: Making methods public just for testing
public class OrderService {
    // Method made public only for testing
    public Boolean isValidOrder(Order__c order) {
        // Validation logic
        return order.Amount__c > 0 && order.Status__c == 'Pending';
    }
    
    public void processOrders() {
        // Process orders using isValidOrder
    }
}

// Better approach: Use @TestVisible
public class OrderService {
    // Private method with @TestVisible annotation
    @TestVisible
    private Boolean isValidOrder(Order__c order) {
        // Validation logic
        return order.Amount__c > 0 && order.Status__c == 'Pending';
    }
    
    public void processOrders() {
        // Process orders using isValidOrder
    }
}

Don’t make methods or variables public just for testing. This exposes implementation details and breaks encapsulation. Instead, use the @TestVisible annotation to make private members accessible to test classes while keeping them private to the rest of your code.

Not Using SObjectType for Field and Object Access

// Anti-pattern: Using string literals for field and object names
public List<SObject> getRecords(String objectName, String fieldName, String value) {
    // Vulnerable to SOQL injection
    String query = 'SELECT Id, Name FROM ' + objectName + 
                   ' WHERE ' + fieldName + ' = \'' + value + '\'';
    return Database.query(query);
}

// Better approach: Use SObjectType for dynamic references
public List<SObject> getRecords(String objectName, String fieldName, String value) {
    // Validate object name
    Schema.SObjectType objType = Schema.getGlobalDescribe().get(objectName);
    if (objType == null) {
        throw new IllegalArgumentException('Invalid object name: ' + objectName);
    }
    
    // Validate field name
    Schema.DescribeSObjectResult objDescribe = objType.getDescribe();
    Map<String, Schema.SObjectField> fieldMap = objDescribe.fields.getMap();
    if (!fieldMap.containsKey(fieldName)) {
        throw new IllegalArgumentException('Invalid field name: ' + fieldName);
    }
    
    // Build and execute query safely
    String safeObjectName = String.valueOf(objType);
    String safeFieldName = String.valueOf(fieldMap.get(fieldName));
    
    String query = 'SELECT Id, Name FROM ' + safeObjectName + 
                   ' WHERE ' + safeFieldName + ' = :value';
    return Database.query(query);
}

Use Schema.SObjectType and Schema.SObjectField for dynamic references to objects and fields instead of string literals. This prevents SOQL injection vulnerabilities and catches errors at compile time rather than runtime.

Not Using WITH SECURITY_ENFORCED

// Anti-pattern: Not enforcing object and field level security
public List<Account> getAccounts() {
    return [SELECT Id, Name, AnnualRevenue FROM Account];
}

// Better approach: Use WITH SECURITY_ENFORCED
public List<Account> getAccounts() {
    return [SELECT Id, Name, AnnualRevenue 
            FROM Account 
            WITH SECURITY_ENFORCED];
}

// Alternative approach: Use Security.stripInaccessible
public List<Account> getAccounts() {
    List<Account> accounts = [SELECT Id, Name, AnnualRevenue FROM Account];
    
    // Strip fields the user doesn't have access to
    SObjectAccessDecision decision = Security.stripInaccessible(
        AccessType.READABLE, accounts
    );
    
    return (List<Account>) decision.getRecords();
}

Enforce object and field level security in your SOQL queries using WITH SECURITY_ENFORCED or Security.stripInaccessible(). This ensures that users can only access data they have permission to see, following the principle of least privilege.

// Anti-pattern: Not specifying sharing settings
public class AccountService {
    public List<Account> getAccounts() {
        return [SELECT Id, Name FROM Account];
    }
}

// Better approach: Explicitly declare sharing model
// With sharing: Respects user's permissions
public with sharing class AccountService {
    public List<Account> getAccounts() {
        return [SELECT Id, Name FROM Account];
    }
}

// Without sharing: Ignores user's permissions (use carefully)
public without sharing class SystemService {
    public void performSystemOperation() {
        // Operations that need system-level access
    }
}

// Inherited sharing: Inherits sharing setting from calling code
public inherited sharing class UtilityService {
    public List<Account> getAccountsWithSharing() {
        return [SELECT Id, Name FROM Account];
    }
}

Always explicitly declare the sharing model for your Apex classes using with sharing, without sharing, or inherited sharing. The with sharing keyword enforces record-level security, while without sharing bypasses it. Use without sharing only when absolutely necessary, and document why it’s needed.

Not Using Proper Trigger Frameworks

// Anti-pattern: Monolithic trigger with business logic
trigger AccountTrigger on Account (before insert, before update, after insert, after update) {
    if (Trigger.isBefore && Trigger.isInsert) {
        // Validation logic
        for (Account acc : Trigger.new) {
            if (acc.Name == null || acc.Name.length() < 3) {
                acc.Name.addError('Account name must be at least 3 characters');
            }
        }
    } else if (Trigger.isAfter && Trigger.isInsert) {
        // Create related records
        List<Contact> contactsToInsert = new List<Contact>();
        for (Account acc : Trigger.new) {
            Contact con = new Contact(
                LastName = 'Primary Contact',
                AccountId = acc.Id
            );
            contactsToInsert.add(con);
        }
        insert contactsToInsert;
    }
    // More logic for other contexts...
}

// Better approach: Use a trigger framework
// Trigger - thin dispatcher
trigger AccountTrigger on Account (before insert, before update, after insert, after update) {
    TriggerHandler.run(new AccountTriggerHandler());
}

// Handler class
public class AccountTriggerHandler extends TriggerHandler {
    public override void beforeInsert() {
        AccountService.validateAccounts(Trigger.new);
    }
    
    public override void afterInsert() {
        AccountService.createPrimaryContacts(Trigger.new);
    }
    
    // Other handler methods...
}

// Service class with business logic
public class AccountService {
    public static void validateAccounts(List<Account> accounts) {
        for (Account acc : accounts) {
            if (acc.Name == null || acc.Name.length() < 3) {
                acc.Name.addError('Account name must be at least 3 characters');
            }
        }
    }
    
    public static void createPrimaryContacts(List<Account> accounts) {
        List<Contact> contactsToInsert = new List<Contact>();
        for (Account acc : accounts) {
            Contact con = new Contact(
                LastName = 'Primary Contact',
                AccountId = acc.Id
            );
            contactsToInsert.add(con);
        }
        insert contactsToInsert;
    }
}

Use a trigger framework to organize your trigger logic. A good framework separates trigger dispatching from business logic, prevents recursive trigger execution, and makes it easier to maintain and test your code. Avoid putting business logic directly in trigger files.

Not Using Proper Naming Conventions

// Anti-pattern: Poor naming conventions
public class cls {
    public void p(List<Account> l) {
        for (Account a : l) {
            a.n = a.n.toUpperCase();
        }
        update l;
    }
}

// Better approach: Use descriptive names
public class AccountService {
    public void processAccounts(List<Account> accounts) {
        for (Account account : accounts) {
            account.Name = account.Name.toUpperCase();
        }
        update accounts;
    }
}

Follow proper naming conventions for classes, methods, and variables. Use descriptive names that clearly indicate the purpose of each entity. This makes your code more readable and maintainable. Class names should be nouns, method names should be verbs, and variable names should be descriptive.

Not Using Asynchronous Apex When Appropriate

// Anti-pattern: Doing too much in synchronous context
public void processLargeDataSet() {
    List<Account> accounts = [SELECT Id FROM Account LIMIT 50000];
    for (Account acc : accounts) {
        // Complex processing
    }
    // More operations that could hit governor limits
}

// Better approach: Use asynchronous Apex
public void processLargeDataSet() {
    // Queue the job for asynchronous processing
    System.enqueueJob(new AccountProcessingJob());
}

// Queueable implementation
public class AccountProcessingJob implements Queueable {
    public void execute(QueueableContext context) {
        // Process accounts in chunks
        List<Account> accounts = [SELECT Id FROM Account LIMIT 1000];
        
        // Process this batch
        for (Account acc : accounts) {
            // Complex processing
        }
        
        // If more accounts to process, chain another job
        List<Account> remainingAccounts = [SELECT Id FROM Account 
                                         WHERE Id > :accounts[accounts.size()-1].Id 
                                         LIMIT 1];
        if (!remainingAccounts.isEmpty()) {
            System.enqueueJob(new AccountProcessingJob());
        }
    }
}

Use asynchronous Apex (Queueable, Batch, Scheduled, or Future methods) for operations that process large amounts of data or might exceed governor limits. Asynchronous Apex has higher governor limits and can be chained or scheduled to process data in manageable chunks.

Getting Started

AI Coding Agents

Axon Models

AI Code Reviews

Integrations

Apex