Hi Samuel, Sorry for the late reply; I have been very busy lately and forgot to get back to you.
I agree that libcurl is the best option for handling HTTP requests. I am already exploring a few possible implementations: 1. Blocking: One curl_easy handle per node performing blocking requests. 2. Non-Blocking: One. curl_easy handle per node managed by a global curl_multi handle for non-blocking IO 3. Custom Library: A dedicated libhttpfs using a ring buffer that wraps libnetfs and exposes HTTP-specific callbacks through libcurl What do you think ? Il giorno mar 27 gen 2026 alle ore 00:47 Samuel Thibault <[email protected]> ha scritto: > > Hello, > > Thinking about these: shouldn't we just use libcurl? I don't see a > reason to maintain http/https code when curl can probably provide just > what we need. > > Samuel
