quick question: could you maybe make it a public bucket, hook it up to a custom domain and then access the files that way? I thought that was a solution for this kind of thing.
well i don't get it then. to download something from a server / serverless function. Something need to be running isn't it? if the fetch request take 0ms, what about the response time to the user? it's not counted in the workers duration cost ?
so @james8313 i believe this means that you'd only be billed for the time it takes the worker to retrieve the data from r2. once it starts responding with the data, the timer stops and you won't be billed for worker time anymore for that request
i guess one way to think about it is that once the worker starts returning the data, the actual worker code isn't running anymore, it's just a data stream
"and a isolate could serve multiple requests at once" really ? i always thought they could serve only 1 request at once and create more instance for every new concurrent request happening at the same time