Pre-orders are live for The Rise of Demons!
Book 1 of The Dark War Trilogy
It’s a dark fantasy mystery rather than full-scale epic fantasy. No chosen ones, no instant world-ending wars. Just a slow unravelling of things the world insists don’t exist.
https://t.co/LHgraQz70k
@CrowderSoup@BenLesh It’s actually more of a situation where both influenced each other. Especially because of Salem.
But there are traditions that evolved and changed because of each other.
We used to carve turnips. America carved pumpkins, so we then switched to carving pumpkins.
@BenLesh I’m shocked because the chances of anyone I know in tech ever making this far up the country is low to non-existent. Missed opportunity for a coffee ☕️
Fair. I’m personally on the fence about both approaches. I can see the simplicity of .subscribe() starting Observable, but I can see why it could become problematic also.
I just feel like there must be some way to compose the observable at construction site that sets out the policies consumers of the observable should adhere to.
But that _does_ increase the complexity and onus on knowing what different values for execution, activation, replay policies actually mean.
This is a tough problem to solve. I appreciate your attention and thoroughness on making sure it’s done right
I’m out of the loop with the entire thread but why not both HotObservable and ColdObservable, putting onus on producer to decide which is needed?
I understand that it means people need to understand the difference between them, but if a default is decided, then whether that default is hot or cold is still impactful to the person using it and they need to:
- adapt their code to handle to handle the potential issues related to one or the other
- reach for an existing library that provides the other form of Observable they want/need.
Hell, the type could carry a Symbol identifying it as Hot or Cold and library maintainers that consume Observable produced externally can branch on the type of observable itself.
Hey everyone!
Supabase CLI V3 just shipped 🚀
There is a migration guide here: https://t.co/OsOrrNaLi8
It should be pretty straightforward to point your agent of choice at the doc and it will migrate you past the breaking changes.
Agent compaction would work better if agents had their own minimal language that they could compress session/context history into, that they could "read" back without decompression.
I’ve been building Rtistree: an artist’s toolkit for agents.
You give direction. Your agent builds, inspects and refines editable artwork.
This demo builds a landscape in layers, then moves just the sun.
Open source · MIT · CLI / MCP / TypeScript SDK
https://t.co/Sv2EQSOV2E
i replaced claude code with oh my pi (omp) as my daily coding agent.
i wrote a guide on how to switch:
• one-command install, log in with your claude max sub
• your skills, slash commands, mcp servers, and CLAUDE.md come with you
• set up safe perm settings (omp defaults to yolo mode)
• set up a different model for each job, across any provider
there's also a prompt you can paste into claude code to do the migration for you.
https://t.co/IK4PaPF1zX