@SandroMaglione I was thinking of building a HttpClientMachine that takes in a HttpClient and infers the machine to be consumed by other machines
What do you think?
Maybe to echo @dax but I disagree.
The affordance to surrender is only true if you let the agent decide the details.
If you decide the details then agents doing the work feels like a speed running exploration.
The final product is rarely the first attempt.
@thdxr Yeah even I'd say at the trade-off of not fulling being able to recreate the callstack from scratch in memory. If the verification is substantial, I don't feel bad not fully mapping out the state machine
@makisuo I find this to be the case as well with products that are being built. Feels like the scope is wide and hard to communicate value without being general
@aboodman What do you suspect is the new skillset? I suspect programming is becoming a supplementary skill and the primary will need to be something else (maybe industry domain knowledge or product decisioning)
@_Shubham2XX To be fair I was stunned by it too. Materialised view in Clickhouse (which I'm assuming is what you're referring to) is fundamentally very different than materialised view in Postgres.
Especially incremental materialized view. It's quite cool the OLAP world.
Interesting, I always thought Effect's scope control felt similar to Rust's (maybe with a much nicer async colouring handling). Glad to hear it translate over well!
I like the `|>` pipe syntax (looks like Elixir). I'm curious how you would evolve the error catching (I didn't see catchTags) and Schema handling
My weekend's now are just rewriting packages that I reuse with native Effect (no dependencies)
I think the complexity of these packages are correctly handled with all of Effect's goodness (will send links when they're done!)
I built a AWS lambda / Planetscale storage adapter for the Workflow/Activity abstraction here
https://t.co/WTOIyoHh43
Working with the effect core team on slack (haven't received much support though). We're trialing it out in our system to make it stable but the examples in the repo should give you a range of what's tested
https://t.co/nq8rPgKys6
OpenAI deactivated my account twice without giving me a reason. The 1st time they admitted it was their fault. The 2nd time they won’t let me appeal.
Timeline:
Jul 28 — Deactivated — C-d1bmXKptYo97
Jul 28 — Appeal submitted
Aug 6 — OpenAI: “We incorrectly deactivated your account.” Restored.
Aug 22 — Deactivated again — C-d1bmXKptYo97
Aug 22 — API access also deactivated — C-LfzKz1IlpJUB (I’ve never created an API key/project)
Aug 22 — Appeal submitted
Aug 22 — OpenAI: “original decision stands.”
Is there anyone from @OpenAI who can help me understand what’s happening? I’m frustrated because I’d happily pay for Pro again, but I’m worried I’ll just lose the account without warning or explanation again.