Joakim L. Christiansen

> a test for this close code should only be run when testing clients, not servers. I'm sorry but this statement doesn't make any sense to me since I was...

Sounds good to me! Such tests can be added at a later time.

I'm not against IPv6 support. But I also don't see any meaning in testing against it if the test over IPv4 already was a success, it's not like anything should...

So D is the private exponent there? I would also really enjoy support for this!

If you do something like this then you should be able to block it yourself before it opens a HTTP connection: ```js const listener = Deno.listen({port, hostname}) for await(const conn...

> BTW, any info how to close connection if use Deno.serve ? I don't even know how to get the IP address related to the requests it responds with or...

Okay. So if doing lots of writes it's best to batch them in an atomic then (to save cost)... Seems weird since initially atomic operations only supported 10 mutations because...

I think it would be better with a pricing system that more directly match our price impact on the underlying FoundationDB (if it could be measured). Because this feels very...

Someone might tell me to just use a subdomain like `api_random_string.my_website.com` but that doesn't really help at all! The bots will still find that server and send requests to it,...

> [teleprint-me:](https://github.com/ggerganov/llama.cpp/issues/5922#issuecomment-1987341505) Yeah, this is why I said templates should be the responsibility of the user. Agreed, which I asked about here [issues/6982](https://github.com/ggerganov/llama.cpp/issues/6982). As ngxson pointed out "the code is...