`stop-when`: the fused token's callback can run twice, or be skipped, when the first token is already stopped
維護者通常 1 天內回覆
評估
- 難度
- 3/5
- 預估耗時
- 1-2 天
- 新手友好度
- 35/100
- Issue 類型
- 缺陷
- 描述清晰度
- 描述清楚
- 活躍度
- 停滯
- 技術堆疊
- cpp
研究方向
The issue identifies __stop_when_::__fused_token<T1, T2>::callback_type and gives a reproducer using stdexec/execution.hpp. Read the callback member declarations and constructor initialization order, then run the reproducer to check immediate invocation when the first token is already stopped. Done means the callback runs once at registration and is not called again when the second token stops. Pull request #2327 is already open against this issue.
由索引模型根據 Issue 內容生成。
描述
Summary
__stop_when_::__fused_token<T1, T2>::callback_type declares its once-flag __called_ after the two inner callbacks:
_Fn __fn_;
__cb1_t __cb1_;
__cb2_t __cb2_;
__std::atomic<bool> __called_{false}; // initialized last
If the first token is already stopped, __cb1_'s constructor invokes the callback immediately. That callback does __called_.exchange(true) on an object whose lifetime hasn't started yet. Then __called_'s initializer stores false. So the result depends on the indeterminate value:
- if it was 0: the function runs at registration, the flag is reset, and the function runs a second time when the second token is stopped;
- if it was non-zero: the function is skipped at registration. It runs only if the second token is stopped later, and is lost otherwise.
Both violate [exec.stop.when]: the callback must be invoked when either token is stopped, and at most once.
Introduced in https://github.com/NVIDIA/stdexec/pull/2234
Reproducer
#include <stdexec/execution.hpp>
#include <cstdio>
namespace ex = STDEXEC;
int main() {
ex::inplace_stop_source a, b;
a.request_stop(); // the sender's token is already stopped
int calls = 0, at_registration = -1;
auto probe = ex::read_env(ex::get_stop_token) | ex::then([&](auto token) {
auto fn = [&]() noexcept { ++calls; };
ex::stop_callback_for_t<decltype(token), decltype(fn)> cb(token, fn); // must run: a is stopped
at_registration = calls;
b.request_stop(); // must not run again
});
// The inner stop-when sees a stoppable receiver token (b), so its child gets a fused token.
ex::sync_wait(ex::__stop_when(ex::__stop_when(probe, a.get_token()), b.get_token()));
std::printf("calls at registration: %d (expected 1), in total: %d (expected 1)\n",
at_registration, calls);
return at_registration == 1 && calls == 1 ? 0 : 1;
}
Then output will be: calls at registration: 1 (expected 1), in total: 2 (expected 1). It should be 1 and 1.
- 主要語言
- C++
- 星號
- 2.5k
- 分支
- 277
- 平均合併
- 1 天 16 小時
- 30 天內合併 PR
- 65
環境準備
這個專案沒有提供開發容器、Dockerfile 或貢獻指南,環境需要你自己搭建:先看它的 README,通用步驟見我們的新手貢獻指南。
從這裡開始
- 先讀完整個 Issue,再讀專案的貢獻指南。
- 在 Issue 下留言說明你要接手 —— 這能避免兩個人做同樣的事。
- Fork 儲存庫,在一個分支上完成修改。
- 送出 Pull Request,並在描述裡引用這個 Issue 編號。
NVIDIA/stdexec 的其他 Issue
-
難度 2/5 1-3 小時 新手友好度 68/100
維護者通常 1 天內回覆
-
難度 5/5 一週以上 新手友好度 12/100
維護者通常 1 天內回覆
-
難度 1/5 1 小時以內 新手友好度 15/100
維護者通常 1 天內回覆
-
難度 4/5 3-5 天 新手友好度 15/100
維護者通常 1 天內回覆
-
難度 3/5 1-2 天 新手友好度 54/100
維護者通常 1 天內回覆
相似的 Issue
-
Mac mini M6 (16 GB): Q2_0 streaming works - measured 11.7->27.2 tok/s + arena fix for sub-32 GB Macs未關閉
難度 2/5 1-3 小時 新手友好度 68/100
維護者通常 1 天內回覆
-
champion good first issue remake source/qq
難度 2/5 1-3 小時 新手友好度 65/100
metasequoiaime/msime-windows#1025 ·
維護者通常 1 天內回覆
-
backlog bug
難度 2/5 1-3 小時 新手友好度 75/100
EchoTools/nevr-runtime#653 ·
維護者通常 1 天內回覆
-
`--fail-require-skip` never fires: run_tests.py searches stdout, but the unittest binary writes the skip summary to stderr可能已有人在做 關聯的 PR 仍在進行中或已合併。 未關閉
難度 2/5 1-3 小時 新手友好度 72/100
維護者通常 1 天內回覆
-
難度 2/5 1-3 小時 新手友好度 75/100
InfiniTimeOrg/InfiniTime#2492 ·