Skip to main content

Idea pipeline (top 25)

Filter by idea status

Filter by product

1967 Ideas

Arie V
Community Manager
Arie VCommunity Manager

IAM: delete authentication details when changing from IAM to External/Windows authentication typeOpen

We are about to move our users from IAM to External authentication type. Testing and reviewing this migration proved very simple: I only have to change the authentication type for the user from IAM to External. This change seems to do nothing more and nothing less than updating the authentication type of the user in the usr_authentication table. The great benefit of SSO is not only a better User Experience, but also improved security, as it eliminates the need for storing User credentials in different systems (i.e. it reduces the ‘attack surface’ for hackers). For the purpose of improved security I would therefore advocate that the (now obsolete) authentication details from the usr_authentication table in IAM should be deleted when changing from IAM to External or Windows authentication type.I intend to run this script myself after having updated the authentication type, but I would rather see IAM doing this by default:UPDATE usr_authentication SET two_factor_authentication_type = 0 , totp_key = null , totp_active = 0 , password_hash = null , password_salt = null , password_algorithm = nullwhere authentication_type = 4 and password_hash is not nullNote: to make sure this doesn't happen accidently or unnoticed, I would recommend adding a warning message or confirmation when an administrator is changing the authentication type from IAM to External.Note 2: I am not certain about Kerberos authentication type, but I think the same would apply.

Add the 'Generated' property for process actions and process variablesClosed

I would like to have the ‘Generated’ property available for process actions and process variables. (Possibly process steps as well since these are dependent on process actions and might cause errors during generation if these aren’t generated as well)I am running into problems with partially generated process flows. A process flow can have the ‘Generated’ property in the SF but individual process actions and process variables cannot. This causes some issues when adding process actions/variables through the dynamic model. To manage these generated process actions and variables, we have to write code to check for them and delete/update them when needed. This is functionality that is available for most other objects in the SF by default.There is also the problem that dynamic model code to manage these object without the ‘Generated’ property can only be executed after the other generated objects are automatically deleted and re-created through the generation process. (Because it is impossible to execute code from the dynamic model before the ‘delete generated objects’ process during generation.) In some cases this causes errors during generation when these objects are dependent on objects that do have the ‘Generated’ property.I’ve noticed that these objects do have a ‘Generation’ group label under ‘Trace info’ but the only column available there is ‘By control procedure’. Does this indicate that the ‘Generated’ property is planned to be added in a later release?

René WSuperhero

Data sets for process testsClosed

In certain situations you have the ability to use process tests or unit tests for testing the same behaviour. As unit tests are more isolated on concepts like defaults etc, in certain situations a process test could be usefull to.The big advantage of unit tests is having Data sets, so mock data can be temporarily inserted in the database. Process tests doesn't have this feature, although I think it would be very usefull there.Process tests are easier to understand: There is no knowledge necessary of ‘logic data’ such as status codes from concepts like context, defaults etc. In stead, building a process test requires only common sense.For example: When you have a subject with a task. And this task may only be enabled when there is a record selected. Testing this can be achieved via a process test using the GUI, like image below.But the same behaviour could be done via unit tests. Only therefore it's necessary to have any knowledge of the ‘expected output’ values, which is very prone to be wrong. The unit test here should be sufficient, but could be more ‘low code’ by having a drop down of options for the expected parameter in some situations. As here the value of the context only can be 0, 1 or 2 (that is, enabled, disabled or hidden).The only problem with the unit test here is you cannot ensure the concept logic is bound to the subject. So the unit test will pass, but you'll still need a process test to ensure the logic is bound to the subject.In many situations the unit test has a big advantage the concept of Data sets is applied here. Is it possible the Data sets also could be applied to a test case or test suite?