A while ago, we posted asking to confirm our visibility of how field visibility works. We’ve been working with this understanding of field visibility. As part of that, if a model is part of multiple datasets and some fields should only be visible in some of those datasets but not all, then we tried to set hidden: true but added the field to the dataset’s custom list of fields. This was working until just recently.
In the past week or so there seems to have been a change: when hidden: true is set, it hides it from the Custom Dataset view as well, even if the field has been added to the dataset custom field list. (Essentially row 1, column 3 in the linked post is now False instead of True) This means a lot of fields that we expected to be visible suddenly are not.
Was this change intentional? This is causing some big headaches given that it’s such a big change from how we confirmed with the holistics team that it works and built our modeling around. If it is intentional, I think we missed any announcement about it.
We’ve checked on our end, and everything appears to be functioning correctly - the fields still display in the Custom Dataset view, even when set as hidden: true. Could you please share the dataset where you’re experiencing this issue by sending it to [email protected]? This will help us investigate further.
We’ve double-checked again and confirmed that this is a bug from our end. The fix has been implemented immediately, and it should now be working. Could you please help check it again?
For context, this regression came from our recent Dataset Field Tree improvements, which gives the users a cleaner hierarchy UI, smoother drag-and-drop experience, better search, and much faster performance on large datasets (announced here).
We missed this case of hidden fields still showing within the Custom Dataset view, and have been actively working on a fix right away after you reported.
Alongside this work, we also identified a bug where metrics with hidden: true are hidden in Custom view. Now they should be shown like other fields, for consistency.
Once again, we sincerely apologize for causing a big headache for your team. If anything still looks off, let us know and we’ll jump on it right away.