Hi!

I understand your problem but I don't think that keep the data in the screen is a good idea. 

Letting the data "alone with the user" is always a invitation to some one try to tamper it, what would ruin the very purpose of your work. 

Perhaps reducing the data to the minimal  and use info already in the server whenever it is possible, but sending the info immediately to the server would be the safer way. 

Just some ideas.

Graça

Hi Joao,

don't have the anwser for you, but here are some thoughts :

# you are looking specifically for a solution for interactions of the end user with the UI, right ?  Any server requests or api calls etc. don't fall under the problem you are having ?

# can you elaborate on what exactly you want to log, is it more tied to the Outsystems structure / logic / bespoke / specific side of your screens, or more to the html / javascript / dom event / generic / technical side.

if it is at a low html / javascript / generic / ... level, maybe you can look at integrating already existing tools to do this, that are not specially geared at Outsystems applications.  

If it is strongly tied to the Outsystems platform or to your specific business needs, then that might not be an option / might be overkill.

# not sure if using client variables would pose a giant performance problem, if you only use it whenever somebody enters or leaves a scope (like a screen, but also maybe at block level)

#Can you share more about what type of data and the amount you are trying to keep track of ?  Is it all moves done by the user, every single click of mouseover, is it all changes to the local data, is it all executions of client actions , ... ?

Dorine

Hi Joao,

depending on your development environment i would suggest the following:

If you have an on-premise installation and access to the SQL Server environment you could make use the SQL Server Audit feature for server and databases and database objects.  See here  https://learn.microsoft.com/en-us/sql/relational-databases/security/auditing/sql-server-audit-database-engine?view=sql-server-ver16 and here https://learn.microsoft.com/en-us/sql/relational-databases/security/auditing/create-a-server-audit-and-database-audit-specification?view=sql-server-ver16

If you have a cloud pipeline you cannot make use of the above. Now in my opinion an audit trail should be 100% accurate. I doubt that you will find any decoupled approach - like the one you describe yourself - that will guarantee that the audit trail is a full representation of the users actions. To make it accurate writing the audit entry must done within the same transaction of the auditable action.

For create, update and delete you would directly add the audit create actions in the flow. This indeed would involve two database roundtrips, but i guess for changes that is ok.

Selecting data can indeed be a problem as you may need to write an audit entry for every resulting record. I never had the requirement to audit list results, but just accessing detail data, so the following is just a way i can think of:

Create server actions to retrieve data using advanced sql. An advanced sql could look like the following

WITH QUERY_CTE
AS
(
  SELECT {DataTable}.[Id],{DataTable}.[Name],{DataTable}.[Other]
  FROM {DataTable}
  WHERE (@SearchTerm = '' OR {DataTable}.[NAME] LIKE '%' + @SearchTerm + '%'
)

INSERT INTO {AuditTrail} (Id,UserId,Table,Name)
SELECT
  NEWID() AS Id,
  @UserId AS UserId,
  'DataTable' AS Table,
  QUERY_CTE.[Name] AS Name
FROM QUERY_CTE

SELECT
  QUERY_CTE.[Id],
  QUERY_CTE.[Name],
  QUERY_CTE.[Other]
FROM QUERY_CTE

Please note that i did not test the above SQL.

Another option is to add an event store to your stack. Like https://www.eventstore.com/

Using an event store you would query your data with an aggregate and then push the result with additional audit data to event store.

Stefan