DML Statements vs Database Methods in Apex (Plus Exception Handling and With Sharing)

Hey friends,

Use a DML statement like insert records; when one failure should stop everything and throw an exception you catch. Use a Database method like Database.insert(records, false) when you want partial success: good records save, bad ones come back as errors in a result list, and no exception is thrown.

PD1 asks you to write DML "and assess what it will return" and to implement exception handling. This follows Trailhead's DML unit.

The second parameter is the whole difference.

DML statements

insert, update, upsert, delete, undelete and merge. If any record fails, the operation throws a DmlException, and control jumps to your catch block.

1try {
2    insert new Account();  // missing required Name
3} catch (DmlException e) {
4    System.debug('A DML exception has occurred: ' + e.getMessage());
5}

Database methods

Each statement has a Database method with an optional allOrNone parameter. Set it to false and the successful records commit while the failures come back in a list of Database.SaveResult. Leave it true and it behaves like the DML statement.

1Database.SaveResult[] results = Database.insert(conList, false);
2for (Database.SaveResult sr : results) {
3    if (!sr.isSuccess()) { for (Database.Error err : sr.getErrors()) { System.debug(err.getMessage()); } }
4}

Trailhead's example: four contacts, one with no LastName. With partial success, three save and one returns an error. To collect the failures, a List or Map helps; see my Apex collections cheat sheet.

Which to use

• All or nothing matters (an order and its lines must save together): DML statement with try/catch.
• Bulk loads where some bad rows are expected: Database method with allOrNone = false, then log or report the failures.
• Either way, never inside a loop. One call on a list. See trigger best practices.

Upsert and merge, quickly

• Upsert inserts new records and updates existing ones in one call, matching on Id or another field you name, which helps avoid duplicates.
• Merge combines up to three records of the same type into one, deletes the others and reparents related records. Only leads, contacts, cases and accounts can be merged.

Exception handling beyond DML

• Catch the specific exception you expect, like DmlException or QueryException, before a generic Exception.
• Create a custom exception by extending Exception, with a name ending in Exception, for your own business errors.
• In triggers, prefer addError() on the record to stop a save with a clean message for the user.

1public class InvoiceException extends Exception {}
2throw new InvoiceException('Invoice total cannot be negative');

With sharing vs without sharing

Declare classes with sharing so the running user's record access applies to the code. Every Trailhead Apex example in these modules does this. without sharing ignores the user's sharing and should be a deliberate, documented choice for a narrow system task. PD1's UI section asks you to prevent data access security vulnerabilities, and this keyword is the first line of defense, along with queries that run in user mode.

Keep studying

• SOQL relationship queries.
• Asynchronous Apex.
• 25 free PD1 practice questions.

Hope this helps!

Best,

Nick