Closed JackKelly-Bellroy closed 2 years ago
Is there some point beyond which we should split hal from event mappings?
It's a good Q; I'm honestly not sure. I think my primary concern with them long term is the lack of independent evolution could lead to noise in breaks--especially given AWS' example-only approach to schemas, which leave them feeling a bit fragile (or overly defensive and less helpful). I at least feel like we're nearing the threshold where we should change the strategy, but I'm not really able to quantify my take here.
Yeah, noise in breaks would be bad. There are several other benefits if we split them out:
aeson
, boot libraries, hedgehog
, hspec
, hspec-hedgehog
, I guess?)Published in 0.4.10! Docs pending. Thanks for another solid addition :)
You're welcome. hal
has been a great library for us, and I'm glad that I can give back.
EventBridge is quite configurable, so it is possible to send a reduced payload or even completely rearrange the JSON that goes onto the event bus. But it's still useful to capture the entirety of an event in a Haskell type.
I'm not sure how many other event mappings I'll need to add from my work, but the scheme I'm establishing here is
AWS.Lambda.Events.EventBridge.#{service}.#{event}
, to let people add events as-needed.Is there some point beyond which we should split
hal
from event mappings?