Oh, one question I had that I forgot to ask - what’s currently the thoughts about automations with triggers other than manual / on new push?
Can you give an example? There is a HTTP API Endpoint for manual so in theory you could plug and play into anything else.
However there is no way of providing per run settings so if you wanted to have something run based on external inputs I am assuming the best way is to use the APIs and push a new commit onto the Model with the external settings apart of that commit?
I think the only thing I can think of off the top of my head is to enable multiple triggers - “whenever A or B updates” but theres also the globals model which would be a good candidate, “whenever project information changes.”
1 Like
Unless I am mistaken, an Automation is assigned per Model so we have worked around that with the SA-Federation Function we created which you would assign to both A and B and point the targetModelName parameter to a Model C which has your actual Automation. Bit messy though
1 Like
I agree this is messy, but as stated in the onboarding, the beta is launched with this limitation clearly stated.
The longer-term roadmap for Automate includes:
- Pipelining/chaining functions within a single automation
- Additional triggers (comments, new models, basically any subset from our webhook triggers we have now)
Automation (small A) should include multiple sophisticated logical interactions. For now, during the beta of Automate (big A), scope out use cases where single functions run on new model data.
2 Likes
I, too, would like a trigger at the project level, with perhaps models being defined as a Function input such that A or B updates would trigger federations and comparisons.
1 Like
How i see things currently:
- “multimodel” automation triggers - totally legit (you want to clash when either x or y changes, for example)
- additional triggers: let’s collect use cases so we don’t spread product wise too thin and loose focus!
incomplete brain dump
To expand on the second one: project info triggers are, in a sense, legit - you'd want to check existing models if models are correctly responding to project info data (e.g, min FAR). Though one could argue the next model version push should detect any incompatibilities too? Anyway, this still holds as a usecase imho.Comments and model creation - I’m a bit wary as we get into automation of “management” tasks and open up questions like how does a user consume outputs from an automation triggered by, for example, a comment (let’s not become jira just yet). Some of these might be actually legimitate “core” product needs, eg. notify model author someone commented, notify project owner new model has been created (+ associated model creation rules/permissions). Pausing here!
3 Likes
Hey folks, thanks for the reply - I’ll accept “lots of great ideas, but outside the scope of the beta” because that’s all I need for my planning right now. Exciting stuff!
2 Likes
Keep them coming - we’re really excited
2 Likes