Stefan Eissing
Stefan Eissing
If there is no further data to add here, I do not see what the curl project can do here. Will close if nothing more is forthcoming.
See my [comment](https://github.com/curl/curl/pull/17246#issuecomment-2969589658) about the PR on how we could tackle this. Propose #17613 to handle this.
> What if we introduce a new value, (`2`) that means what when this transfer starts, it _also_ marks all currently alive connections as _not-for-reuse_ ? Because this functionality is...
> This looks good, but we need this properly documented as well! I do not think closing all existing connection is what we want. A "network change" does not *have...
the early data test failed with other PRs sometimes as well. Seems I need to check what is going on there.
The environment variable were dropped since we could not find a reliable way to make them work on Windows. The env created for the subprocess was missing vital variables needed...
Unless someone makes the code that adds environment variables to mod_md scripts that preserves all existing env vars (which might be possible but I failed to achieve), breaking existing deployments...
No one steps forward, closing.
As with #13288, this seems to call for further clarification on how websocket transfers should handle non-blocking `EAGAIN` situations. And how that should work in a multi or standalone easy...
I just made improvements in `docs/examples/websocket.c` to show how to handle `CURLE_AGAIN` on send/receive of websockets. That should illustrate this better. As for running a websocket connection via `CURLOPT_WRITEFUNCTION` and...