core.net — Networking

Import with << core.net. See the corelib index.

@tcpRequest, the raw TCP request-exchange primitive the HTTP client sits on.

Function Effect

@tcpRequest(address :: Text, requestBytes :: Text) -> Result One-shot request exchange: connect to address (host:port), write requestBytes, read the response until the peer closes (close-delimited). Yields Ok(responseBytes) with the whole response as a Text on success, or NotOk(errorMessage) on ANY network failure (DNS resolution, connect, write, or read) — a failure is a value to match, never a crash. A value-returning leaf IO primitive: the call launches the exchange and hands back a deferred Result, forced when a strict operation first reads it.

Function	Effect
`@tcpRequest(address :: Text, requestBytes :: Text) -> Result`	One-shot request exchange: connect to `address` (`host:port`), write `requestBytes`, read the response until the peer closes (close-delimited). Yields `Ok(responseBytes)` with the whole response as a `Text` on success, or `NotOk(errorMessage)` on ANY network failure (DNS resolution, connect, write, or read) — a failure is a value to match, never a crash. A value-returning leaf IO primitive: the call launches the exchange and hands back a deferred `Result`, forced when a strict operation first reads it.

The response is capped at 16 MiB; a larger one yields NotOk rather than exhausting memory. Hostname resolution is a blocking DNS lookup on the fiber thread, so a slow lookup stalls the scheduler (a numeric host:port skips it); non-blocking DNS is a later refinement.

<< core.net
<< core.io

reportFailure = (message :: Text) -> Num => <
  eprint(message)   ~ error to stderr…
  1                 ~ …and a non-zero exit
>

^ = () -> Num =>
  @tcpRequest("localhost:8080", "GET / HTTP/1.0\r\n\r\n") ?
    | Ok(response) => response.size > 0 ? 0 : 1   ~ forced by the match
    | NotOk(error) => reportFailure(error)

Being deferred, independent requests on one fiber overlap automatically — each forces where its outcome is first read. See the Concurrency model, and examples/net_request.qn for a real HTTP GET over core.net.