Content
@
0 reply
0 recast
2 reactions
Shane da Silva
@sds
A new year's resolution is to write more technical blog posts about challenges we're working on for Warpcast + Farcaster. Here's a post discussing our migration from Faktory to our own hand-rolled job processing framework, and the lessons learned. https://paragraph.xyz/@sds.eth/building-a-job-processing-system
9 replies
113 recasts
254 reactions
Alex Loukissas 🍉
@futureartist
Did you consider Temporal? Also in OSS land, this is a good one to take inspo from (we're happily using it in production, only dependency is postgres) https://github.com/oban-bg/oban
1 reply
1 recast
1 reaction
Shane da Silva
@sds
We use Temporal for a subset of high-value jobs (read: money processing), but it felt like overkill for all jobs, especially at the volume we were doing. If you really care about what the exact state of a job is and it’s not an inherently idempotent operation that can simply be retried, Temporal seems to be a reliable choice. Haven’t heard of oban—thanks for putting it on the radar. In our case, it’s really nice to have the job framework also written in TypeScript, since our entire backend codebase is TypeScript.
1 reply
0 recast
1 reaction
Alex Loukissas 🍉
@futureartist
Oh absolutely. I just shared for potential code reading to take ideas. It’s a really mature project. So it may save you time in the future- maybe!
1 reply
0 recast
2 reactions
Samuel ツ
@samuellhuber.eth
In the installation it seems to be elixir only? Or can one use it with TS
2 replies
0 recast
0 reaction
Alex Loukissas 🍉
@futureartist
Like imagine you deploy it as a black-box elixir app with docker and use it from nodejs, golang, etc.
1 reply
0 recast
2 reactions
Alex Loukissas 🍉
@futureartist
Well yeah I think the interface and the engine is elixir only. But! It may be interesting to expose e.g. grpc/http interface to it. The engine and the admin UI are so well made.
0 reply
0 recast
1 reaction
