i find these discussions comparing the "vision language models" to the old computer vision tech pretty interesting
since there are still strengths the computer vision has, i wonder why someone hasn't made an "über vision language service" that just exposes the old CV APIs as MCP or something, and have both systems work in conjunction to increase accuracy and understanding
I still use it for that. Built a really nice packer based process that pulls in Chef cookbooks and applies them in zero mode. Even have an integration test of the resulting AMI in later stages using test kitchen (Chef's "killer app" imo) with the EC2 driver to spin up an instance with the AMI and run Inspec tests on it.
There are many types of request that cannot be made with client-side JS alone, but for those that can, the ability to send those requests client-side would be handy.
I think that 99.9% of CURL commands are copied from Chrome/Firefox's network inspector and are the simple "client-side JS" types.
I also think it's weird to be so willing to let people run arbitrary CURL commands from your platform, without any billing or account verification. It feels ripe for abuse.
Even for commands that use the subset of cURL features that fetch supports, most requests to other domains (cross origin requests) wouldn't work anyway because the responses won't have the CORS[0] headers to allow being accessed from arbitrary websites. So running it client side would be infeasible for most requests.
Exactly — this was one of the core constraints I ran into early on.
CORS was a blocker for client side requests, I have a separate branch where this is integrated, maybe will add it alongside server side execution to let the person creating the curl decide whether they can execute on browser or server side.
reply