Files
jak-project/third-party/curl/tests/http/test_08_caddy.py
T
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

223 lines
9.8 KiB
Python
Vendored
Generated

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
#***************************************************************************
# _ _ ____ _
# Project ___| | | | _ \| |
# / __| | | | |_) | |
# | (__| |_| | _ <| |___
# \___|\___/|_| \_\_____|
#
# Copyright (C) Daniel Stenberg, <daniel@haxx.se>, et al.
#
# This software is licensed as described in the file COPYING, which
# you should have received as part of this distribution. The terms
# are also available at https://curl.se/docs/copyright.html.
#
# You may opt to use, copy, modify, merge, publish, distribute and/or sell
# copies of the Software, and permit persons to whom the Software is
# furnished to do so, under the terms of the COPYING file.
#
# This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
# KIND, either express or implied.
#
# SPDX-License-Identifier: curl
#
###########################################################################
#
import difflib
import filecmp
import logging
import os
import re
import pytest
from testenv import Caddy, CurlClient, Env, LocalClient
log = logging.getLogger(__name__)
@pytest.mark.skipif(condition=not Env.has_caddy(), reason="missing caddy")
@pytest.mark.skipif(condition=not Env.have_ssl_curl(), reason="curl without SSL")
class TestCaddy:
@pytest.fixture(autouse=True, scope='class')
def caddy(self, env):
caddy = Caddy(env=env)
assert caddy.initial_start()
yield caddy
caddy.stop()
def _make_docs_file(self, docs_dir: str, fname: str, fsize: int):
fpath = os.path.join(docs_dir, fname)
data1k = 1024*'x'
flen = 0
with open(fpath, 'w') as fd:
while flen < fsize:
fd.write(data1k)
flen += len(data1k)
return flen
@pytest.fixture(autouse=True, scope='class')
def _class_scope(self, env, caddy):
self._make_docs_file(docs_dir=caddy.docs_dir, fname='data10k.data', fsize=10 * 1024)
self._make_docs_file(docs_dir=caddy.docs_dir, fname='data1.data', fsize=1024 * 1024)
self._make_docs_file(docs_dir=caddy.docs_dir, fname='data5.data', fsize=5 * 1024 * 1024)
self._make_docs_file(docs_dir=caddy.docs_dir, fname='data10.data', fsize=10 * 1024 * 1024)
self._make_docs_file(docs_dir=caddy.docs_dir, fname='data100.data', fsize=100 * 1024 * 1024)
env.make_data_file(indir=env.gen_dir, fname="data-10m", fsize=10 * 1024 * 1024)
# download 1 file
@pytest.mark.parametrize("proto", Env.http_protos())
def test_08_01_download_1(self, env: Env, caddy: Caddy, proto):
curl = CurlClient(env=env)
url = f'https://{env.domain1}:{caddy.port}/data.json'
r = curl.http_download(urls=[url], alpn_proto=proto)
r.check_response(count=1, http_status=200)
# download 1MB files sequentially
@pytest.mark.parametrize("proto", Env.http_protos())
def test_08_02_download_1mb_sequential(self, env: Env, caddy: Caddy, proto):
count = 50
curl = CurlClient(env=env)
urln = f'https://{env.domain1}:{caddy.port}/data1.data?[0-{count-1}]'
r = curl.http_download(urls=[urln], alpn_proto=proto)
r.check_response(count=count, http_status=200, connect_count=1)
# download 1MB files parallel
@pytest.mark.parametrize("proto", Env.http_protos())
def test_08_03_download_1mb_parallel(self, env: Env, caddy: Caddy, proto):
count = 20
curl = CurlClient(env=env)
urln = f'https://{env.domain1}:{caddy.port}/data1.data?[0-{count-1}]'
r = curl.http_download(urls=[urln], alpn_proto=proto, extra_args=[
'--parallel'
])
r.check_response(count=count, http_status=200)
if proto == 'http/1.1':
# http/1.1 parallel transfers will open multiple connections
assert r.total_connects > 1, r.dump_logs()
else:
assert r.total_connects == 1, r.dump_logs()
# download 5MB files sequentially
@pytest.mark.skipif(condition=Env().slow_network, reason="not suitable for slow network tests")
@pytest.mark.parametrize("proto", Env.http_mplx_protos())
def test_08_04a_download_10mb_sequential(self, env: Env, caddy: Caddy, proto):
count = 40
curl = CurlClient(env=env)
urln = f'https://{env.domain1}:{caddy.port}/data5.data?[0-{count-1}]'
r = curl.http_download(urls=[urln], alpn_proto=proto)
r.check_response(count=count, http_status=200, connect_count=1)
# download 10MB files sequentially
@pytest.mark.skipif(condition=Env().slow_network, reason="not suitable for slow network tests")
@pytest.mark.parametrize("proto", Env.http_mplx_protos())
def test_08_04b_download_10mb_sequential(self, env: Env, caddy: Caddy, proto):
count = 20
curl = CurlClient(env=env)
urln = f'https://{env.domain1}:{caddy.port}/data10.data?[0-{count-1}]'
r = curl.http_download(urls=[urln], alpn_proto=proto)
r.check_response(count=count, http_status=200, connect_count=1)
# download 10MB files parallel
@pytest.mark.skipif(condition=Env().slow_network, reason="not suitable for slow network tests")
@pytest.mark.parametrize("proto", Env.http_protos())
def test_08_05_download_1mb_parallel(self, env: Env, caddy: Caddy, proto):
if proto == 'http/1.1' and env.curl_uses_lib('mbedtls'):
pytest.skip("mbedtls 3.6.0 fails on 50 connections with: "
"ssl_handshake returned: (-0x7F00) SSL - Memory allocation failed")
count = 50
curl = CurlClient(env=env)
urln = f'https://{env.domain1}:{caddy.port}/data10.data?[0-{count-1}]'
r = curl.http_download(urls=[urln], alpn_proto=proto, extra_args=[
'--parallel'
])
r.check_response(count=count, http_status=200)
if proto == 'http/1.1':
# http/1.1 parallel transfers will open multiple connections
assert r.total_connects > 1, r.dump_logs()
else:
assert r.total_connects == 1, r.dump_logs()
# post data parallel, check that they were echoed
@pytest.mark.parametrize("proto", Env.http_protos())
def test_08_06_post_parallel(self, env: Env, httpd, caddy, proto):
# limit since we use a separate connection in h1
count = 20
data = '0123456789'
curl = CurlClient(env=env)
url = f'https://{env.domain2}:{caddy.port}/curltest/echo?id=[0-{count-1}]'
r = curl.http_upload(urls=[url], data=data, alpn_proto=proto,
extra_args=['--parallel'])
r.check_stats(count=count, http_status=200, exitcode=0)
for i in range(count):
with open(curl.response_file(i)) as fr:
respdata = fr.readlines()
assert respdata == [data]
# put large file, check that they length were echoed
@pytest.mark.parametrize("proto", Env.http_protos())
def test_08_07_put_large(self, env: Env, httpd, caddy, proto):
# limit since we use a separate connection in h1<
count = 1
fdata = os.path.join(env.gen_dir, 'data-10m')
curl = CurlClient(env=env)
url = f'https://{env.domain2}:{caddy.port}/curltest/put?id=[0-{count-1}]'
r = curl.http_put(urls=[url], fdata=fdata, alpn_proto=proto)
exp_data = [f'{os.path.getsize(fdata)}']
r.check_response(count=count, http_status=200)
for i in range(count):
with open(curl.response_file(i)) as fr:
respdata = fr.readlines()
assert respdata == exp_data
@pytest.mark.parametrize("proto", Env.http_protos())
def test_08_08_earlydata(self, env: Env, httpd, caddy, proto):
if not env.curl_can_early_data():
pytest.skip('TLS earlydata not implemented')
if proto == 'h3' and not env.curl_can_h3_early_data():
pytest.skip("h3 early data not supported")
count = 2
docname = 'data10k.data'
url = f'https://{env.domain1}:{caddy.port}/{docname}'
client = LocalClient(name='cli_hx_download', env=env)
if not client.exists():
pytest.skip(f'example client not built: {client.name}')
r = client.run(args=[
'-n', f'{count}',
'-C', env.ca.cert_file,
'-e', # use TLS earlydata
'-f', # forbid reuse of connections
'-r', f'{env.domain1}:{caddy.port}:127.0.0.1',
'-V', proto, url
])
r.check_exit_code(0)
srcfile = os.path.join(caddy.docs_dir, docname)
self.check_downloads(client, srcfile, count)
earlydata = {}
for line in r.trace_lines:
m = re.match(r'^\[t-(\d+)] EarlyData: (-?\d+)', line)
if m:
earlydata[int(m.group(1))] = int(m.group(2))
assert earlydata[0] == 0, f'{earlydata}'
if proto == 'h3':
assert earlydata[1] == 113, f'{earlydata}'
else:
# Caddy does not support early data on TCP
assert earlydata[1] == 0, f'{earlydata}'
def check_downloads(self, client, srcfile: str, count: int,
complete: bool = True):
for i in range(count):
dfile = client.download_file(i)
assert os.path.exists(dfile)
if complete and not filecmp.cmp(srcfile, dfile, shallow=False):
with open(srcfile) as fa, open(dfile) as fb:
a = fa.readlines()
b = fb.readlines()
diff = "".join(difflib.unified_diff(a=a, b=b,
fromfile=srcfile,
tofile=dfile,
n=1))
assert False, f'download {dfile} differs:\n{diff}'