very quick q: is flow version group actually deprecated like https://docs.prefect.io/orchestration/concepts/flows.html#versioning says? I don't see any references in the code to this deprecation. my use case is: I'd like to split out the "staging" flow runs (which run many times per day but are not very important) from the "prod" flow runs in the UI somehow, and

version_group_id

seems like the most appropriate way to do that...but I don't want to rely on it if it might go away

So this is a planned deprecation but it’s very low priority right now; even if we deprecated it, the graphQL schema would remain stable for quite some time before it would become a breaking change for you. So while I don’t want to “recommend” you use

version_group_id

, I think it’s safe and ultimately the change will be minimal once it is updated (most likely

version_group_id -> flow_group_id

). The reason for the planned deprecation is that version groups and flow groups are somewhat ambiguous and overlapping concepts that we want to solidify into a single interface. Track / comment on this issue to stay up to date: https://github.com/PrefectHQ/server/issues/122

+1 to it being confusing 😅 if the UI had a bit more filtering logic I don't think I would care at all about this anymore, it's just one means to an end. basically just want this plot to omit all but the 2 tallest bars which are our prod runs (and hence take hours instead of minutes); the the others are CI runs on mock data

interesting - so filtering by label / name would possibly even work? cc: @nicholas

probably, though right now we only have one agent that runs everything so only one label. but that'd be easy to change if needed. not sure about name since it's all the same flow but I could change the run names to whatever if that was an option

My thought is name is probably not super useful without a lot of planning ahead of time but filtering by label is definitely something we could work into some upcoming changes to that chart 🙂

yea you can change run names in the UI, and also there’s an API route for it that you could call from within the run itself (I think there’s even a task in the task library for this, but if it were me I’d use a state handler at the flow level)

we do set the run names already, I just don't think they're really amenable to filtering programmatically (but they could be)