Files
jak-project/third-party/curl/docs/libcurl/curl_ws_recv.md
Alexander J. Semenuk 5d5e35fb9b fix: Windows toolchain compatibility (curl 8.21 re-vendor, endless reconfigure loop) (#4355)
## Problem

Windows builds break with a current local toolchain (Scoop LLVM 22.1.8,
CMake 4.4.0, VS 2026), in two independent ways:

1. The build stops at curl's deliberate guard: `#error "no non-blocking
method was found/used/set"` in `third-party/curl/lib/nonblock.c`.
2. From the second configure onward, `cmake --build` re-runs CMake in an
endless loop (observed 42 consecutive reconfigure cycles in a single
build). Likely the same mechanism behind the "endlessly building" VS
2026 note in `docs/setup/dev/vs.md`.

## Root cause

1. `third-party/curl/CMake/CurlTests.c` passes `int *` to
`ioctlsocket()`, whose third parameter is `u_long *`. Clang 22 promotes
`-Wincompatible-pointer-types` to a hard error in C, so the
`HAVE_IOCTLSOCKET_FIONBIO` try_compile silently fails and
`curl_config.h` never defines it. Upstream CI does not see this because
the windows-2022 runner image ships an older LLVM. GCC 14 promotes the
same warning to a hard error, which is very likely the `CurlTests.c.obj`
failure reported from MSYS2 in open-goal/jak-project#3551. Upstream curl
hit the identical problem with GCC 14 and fixed the probe in curl 8.8.0
(curl/curl#13578).
2. The root CMakeLists copies the build tree's `compile_commands.json`
into `<src>/build/` for clangd using `configure_file()`, which registers
its input as a configure dependency. CMake rewrites
`compile_commands.json` late in every generation, after
`CTestTestfile.cmake` and `cmake_install.cmake` (outputs of the same
Ninja regen rule), so once the dependency is registered the rule is
deterministically dirty and every `ninja` invocation re-runs CMake. A
pristine first configure is safe (the file does not exist yet, so the
`if(EXISTS ...)` guard skips the copy), which is why the loop looks
machine- or IDE-specific.

## Fix

1. Per review, re-vendor `third-party/curl` at the `curl-8_21_0` tag
(previously `curl-8_3_0`), which carries the upstream probe fix plus two
years of upstream development; `vendor.yaml` updated to match.
Adjustments the version jump forced:
- curl 8.15 removed the native macOS Secure Transport backend
(`CURL_USE_SECTRANSP`), so macOS now builds curl against OpenSSL like
Linux. The two macOS workflows install Homebrew `openssl@3` and export
`OPENSSL_ROOT_DIR` (keg-only), and the macOS setup docs gained the same
two lines.
- `CURL_BROTLI` / `CURL_ZSTD` switched to AUTO-detection in curl 8.10;
pinned OFF to keep the previous no-compression behavior and avoid
silently linking whatever the CI images happen to have.
- curl's new top-level `BUILD_EXAMPLES` cache option (default ON) leaked
into discord-rpc's identically named option and broke configure at a
nonexistent `examples/send-presence` directory; pinned OFF ahead of the
third-party subdirectories.

The diff is dominated by the mechanical tag-tree swap under
`third-party/curl` (linguist-vendored, collapsed in review). The
hand-written changes are `CMakeLists.txt`, the two macOS workflows,
`docs/setup/system/macos.md`, and `vendor.yaml`.
2. Swap `configure_file()` for `file(COPY ...)`: the same clangd copy
with no configure dependency registered. (`file(COPY_FILE ...
ONLY_IF_DIFFERENT)` would be cleaner still but requires CMake 3.21,
above the declared `cmake_minimum_required(VERSION 3.10)`.)

## Test plan

- [x] Fresh `cmake --preset Release-windows-clang` (LLVM 22, no cache
seeding) completes and logs `Enabled SSL backends: Schannel`; the
FIONBIO probe passes without the previous `#error`
- [x] Full Windows Release build from scratch in the branch worktree
(all 1422 targets)
- [x] goalc-test suite: 1509 passed, 0 failed
- [x] Second consecutive configure with `compile_commands.json` present:
the regen rule in `build.ninja` has no `compile_commands.json` input;
`<src>/build/compile_commands.json` is still refreshed for clangd
- [x] Repeated `ninja` invocations after a full build no longer re-run
CMake
- [x] macOS Intel and ARM CI green (first exercise of the OpenSSL
backend switch)

---

I work off a self-hosted forge, so this GitHub account is quiet; the
configure logs and ninja dirty-node traces from the investigation are
available if anyone wants the raw data.

(AI-assisted)
2026-07-27 19:19:18 -04:00

4.3 KiB
Vendored
Generated

c, SPDX-License-Identifier, Title, Section, Source, See-also, Protocol, Added-in
c SPDX-License-Identifier Title Section Source See-also Protocol Added-in
Copyright (C) Daniel Stenberg, <daniel@haxx.se>, et al. curl curl_ws_recv 3 libcurl
curl_easy_getinfo (3)
curl_easy_perform (3)
curl_easy_setopt (3)
curl_ws_send (3)
libcurl-ws (3)
WS
7.86.0

NAME

curl_ws_recv - receive WebSocket data

SYNOPSIS

#include <curl/curl.h>

CURLcode curl_ws_recv(CURL *curl, void *buffer, size_t buflen,
                      size_t *recv, const struct curl_ws_frame **meta);

DESCRIPTION

Retrieves as much as possible of a received WebSocket frame into the buffer, but not more than buflen bytes. recv is set to the number of bytes actually stored.

If the function call is successful, the meta pointer gets set to point to a const struct curl_ws_frame that contains information about the received data. That struct must not be freed and its contents must not be relied upon anymore once another WebSocket function is called. See curl_ws_meta(3) for more details on that struct.

The application must check meta->bytesleft to determine whether the complete frame has been received. If more payload is pending, the application must call this function again with an updated buffer and buflen to resume receiving. This may for example happen when the data does not fit into the provided buffer or when not all frame data has been delivered over the network yet.

If the application wants to read the metadata without consuming any payload, it may call this function with a buflen of zero. Setting buffer to a NULL pointer is permitted in this case. Note that frames without payload are consumed by this action.

If the received message consists of multiple fragments, the CURLWS_CONT bit is set in all frames except the final one. The appropriate CURLWS_TEXT or CURLWS_BINARY flag is set in every frame, regardless whether it is the first fragment, an intermediate fragment or the final fragment. The application is responsible for reassembling fragmented messages. Special care must be taken to correctly handle control frames (i.e. CLOSE, PING and PONG) arriving in between consecutive fragments of a fragmented TEXT or BINARY message. See curl_ws_meta(3) for more details on CURLWS_CONT.

The WebSocket protocol consists of messages that can be delivered over the wire as one or more frames - but since a frame can be too large to buffer in memory, libcurl may need to deliver partial frames to the application. Fragments, or chunks, of frames.

%PROTOCOLS%

EXAMPLE

int main(void)
{
  char buffer[256];
  size_t offset = 0;
  CURLcode result = CURLE_OK;
  CURL *curl = curl_easy_init();

  curl_easy_setopt(curl, CURLOPT_URL, "wss://example.com/");
  curl_easy_setopt(curl, CURLOPT_CONNECT_ONLY, 2L);
  /* start HTTPS connection and upgrade to WSS, then return control */
  curl_easy_perform(curl);

  /* Note: This example neglects fragmented messages.
     (CURLWS_CONT bit) A real application must handle them
     appropriately. */

  while(!result) {
    size_t recv;
    const struct curl_ws_frame *meta;
    result = curl_ws_recv(curl, buffer + offset,
                          sizeof(buffer) - offset, &recv,
                          &meta);
    offset += recv;

    if(result == CURLE_OK) {
      if(meta->bytesleft == 0)
        break; /* finished receiving */
      if(meta->bytesleft > (curl_off_t)(sizeof(buffer) - offset))
        result = CURLE_TOO_LARGE;
    }

    if(result == CURLE_AGAIN)
      /* in real application: wait for socket here, e.g. using select() */
      result = CURLE_OK;
  }

  curl_easy_cleanup(curl);
  return (int)result;
}

%AVAILABILITY%

RETURN VALUE

This function returns a CURLcode indicating success or error.

CURLE_OK (0) means everything was OK, non-zero means an error occurred, see libcurl-errors(3). If CURLOPT_ERRORBUFFER(3) was set with curl_easy_setopt(3) there can be an error message stored in the error buffer when non-zero is returned.

Returns CURLE_GOT_NOTHING if the associated connection is closed.

Instead of blocking, the function returns CURLE_AGAIN. The correct behavior is then to wait for the socket to signal readability before calling this function again.

Any other non-zero return value indicates an error. See the libcurl-errors(3) man page for the full list with descriptions.

Returns CURLE_GOT_NOTHING if the associated connection is closed.