First of all, historical runs of dataflows should be saved in the monitor section. If you re-run a dataflow, you lose the previous run in the monitor section. That is absolutely horrid, all runs for dataflows should be documented and saved.
The errors themselves and how you navigate to them are also bad. You should be able to open the error UI from the monitor section just like everything else. You shouldn't have to go to the workspace then hover over a little red indicator to see the errors, that is piss poor UI design.
Let me give you a scenario where the lack of detail in error messages leads to problems: you made a table on the warehouse/lakehouse side using CREATE TABLE so that you have precise SQL types. You go back to the dataflow and all the column mappings lineup BUT, one of the VARCHAR lengths is too short for the items in that column. There isn't any errors so you go run the dataflow and it takes ages, then returns an error. But guess what, it doesn't tell you which column or why it failed.
I spent days going in circles only to query in the information schema directly to my ERP to finally notice that I had items that were larger than the VARCHAR length I had set for that table. The fact the errors can't immediately direct me to this is an indicator of how far behind the underlying dataflow infrastructure is compared to other items in Fabric. This has got to improve!
