Forum Discussion
Mark
Thanks for quick reply. I actually have had some help from Don in setting this app up. He helped me set up what you mention above (Max RID#) and I have a button in my REQUEST DECISIONS report that allows an approver to make their decision via the button, which saves the record in DECISION STATUS CHANGES table that is pulled into REQUEST DECISIONS via Max RID.
Button formula: If([LEAVE REQUEST - LEAVE STATUS]="FINAL APPROVAL", "", URLRoot() & "db/" & [_DBID_DECISION_STATUS_CHANGES] & "?a=API_GenAddRecordForm&_fid_26=" & URLEncode ([Record ID#])& "&dfid=11" & "z=" & Rurl())
So you're saying that you would go with the opposite recommendation from chatgpt?
I guess what I'm saying is that independently Don and myself had the same recommended solution. Your source of truth is the Child table and you flow the most recent status update up to the parent record using formulas and summary fields which can never fail. So yes, imho, our two brains are smarter than Chat GPT.
Under the Chat GPT recommended solution you can have a theoretical mismatch between the children and the parent record status, then you might need to have a safety net report to detect those differences and manually correct them.