Apex Code Development You need to sign in to do that
Don't have an account?
Paul Weeden New Releases Breaking Code
Dear Salesforce,
Every time a new version of the platform is about to be released the entire system becomes hugely unreliable.
I have some managed code that I tested last night, early this morning, early this afternoon and late this afternoon. The test methods have not changed and we have not changed the configuration of the environment. Last night the tests worked, this morning they failed, early afternoon they succeed and this late this afternoon they suddenly stopped working again!
I've just tried to edit a record that passed validation previously and it is now failing. The validation is failing with some code that checks for two non-null values. If either value is not null an error is thrown. Trouble is, both values are null so the trigger should never get to that part of the code!
This is failing in the development org as well as our client's subscriber orgs.
Not a great advert for either of us!
Please can you help.
Thanks
I hope you are sure that there no issue regarding data on your org, as there are many processes like workflow , approval etc which may change the data.
Yep. When I went hom last night I clicked 'Run All Tests'. Came in this morning to see test failures in one of the classes. Thought it was strange, so clicked it again and all passed! Managed to deploy the package this afternoon, installed in subscriber org OK. Created a few test records in subscriber org's sandbox (same package version) then went to demo it and received the error.
It's very strange. Normally I'd accept it was probably something we've done, but given the very strange behaviour we've seen before in the lead up to new releases AND the funny test failures I wanted to get some feedback from the community.
Since writing the 1st message I'm now getting a different error message - still from our code - but in a different section. The record values haven't changed, neither has the code (except for some debug output).
I think your test methods could be organization data dependent. Try to run failed methods seperately and check whether they fail or not.
I can assure you they are not. I have just deleted every record in the org as we ran out of space and to ensure there weren't any hardcoded Ids.
I'm not suggesting that it is deliberate, but when a test succeeds, fails, succeeds, succeeds and then fails all with no underlying code changes then it would suggest an external factor.