跳转至内容
  • A place to talk about whatever you want

    36 主题
    184 帖子
    modou12138-rgbM

    There is a noticeable difference between entering an encounter with a prepared character and entering one hoping that equipment alone will carry the fight. Aion 2 rewards players who understand their abilities, pay attention to enemy behavior, and build characters that can handle the situations they regularly face. Equipment plays a major role in preparation, but the best choices are not always the most obvious ones. I usually consider Aion 2 Kinah part of a larger progression budget, rather than something to spend whenever a new piece of gear catches my attention.
    Preparation starts with knowing what kind of challenge lies ahead. Before attempting difficult content, think about your character's current strengths and weaknesses. Are you comfortable with your combat rotation? Can you react quickly when enemies change their attack patterns? Does your equipment support the way you normally approach fights? These questions are worth asking because an encounter can expose weaknesses that are easy to overlook during ordinary gameplay. Sometimes the solution is an equipment improvement, while other times it is simply becoming more familiar with the fight.
    I particularly enjoy the process of adjusting a character after a difficult encounter. Instead of immediately assuming that the equipment is insufficient, look at what happened during the fight. Maybe you used important abilities at an inconvenient moment, positioned yourself poorly, or failed to recognize a dangerous attack. Reviewing those mistakes can reveal improvements that cost nothing. When an equipment change is genuinely needed, you will have a much better understanding of what to prioritize.
    The community side of equipment progression deserves attention as well. Comparing builds and discussing different item choices can introduce players to approaches they might not have considered. However, I would avoid copying another player's entire setup without understanding the reasoning behind it. Different classes, combat preferences, and progression stages can produce very different equipment priorities. A useful discussion should explain how an item contributes to a particular build instead of simply declaring that one option is always superior.
    Players also need to think about how much time they can realistically invest in collecting and improving equipment. Some enjoy the gradual process of earning everything through regular gameplay, while others want to spend more time on the content they find most entertaining. When researching additional acquisition options, I would pay attention to the service description, pricing, availability, and any relevant game rules. U4GM provides an organized place to review item-related options, which can be useful when comparing different possibilities before making a decision.
    What makes equipment progression interesting is that there is rarely one perfect answer for every character. A useful upgrade for one player might be unnecessary for another, and a carefully developed build can sometimes outperform a more expensive setup that lacks proper planning. I would rather see players discuss how particular equipment choices affect their actual gameplay than focus entirely on displayed values. For anyone considering additional ways to obtain equipment, Aion 2 Items for sale is a topic worth exploring while checking the available options and deciding which ones genuinely match their character's needs.

  • 14 主题
    28 帖子
    modou12138-rgbM

    One of the most enjoyable parts of MyTEAM is seeing a squad gradually take shape through a series of decisions. The first lineup might have obvious weaknesses, but after several upgrades and adjustments, the team starts to feel more personal. You begin to understand which players fit your preferred pace, which matchups cause trouble, and what kind of offensive setup gives you the most consistent results. Managing 2K27 MT can support that process, especially when you have a clear idea of what you want your team to become.
    I think setting short-term roster goals is more useful than trying to improve every position at once. For example, you might decide that your next priority is finding a dependable center who can protect the paint and secure rebounds. Once that position is covered, you can turn your attention to perimeter defense or outside shooting. This approach makes progress easier to measure and prevents resources from disappearing through several unrelated purchases.
    There is also a difference between building a team for individual player strengths and building one around an actual offensive system. A talented scorer may be excellent in isolation, but that does not mean every possession should revolve around him. Other players need opportunities to contribute through screens, cuts, passing, and offensive rebounds. When the entire lineup has a useful role, opponents have more decisions to make and fewer opportunities to concentrate on stopping one player.
    Learning different defensive approaches is equally valuable. Some opponents prefer driving into the paint, while others rely heavily on perimeter shooting or repeated pick-and-roll actions. Changing defensive assignments, paying attention to help defense, and avoiding unnecessary reaching can make a noticeable difference. I would rather force an opponent into a difficult attempt than constantly gamble for a steal and leave an open player behind me.
    Player familiarity is another factor that should not be underestimated. Switching cards too frequently can make it harder to develop consistent shooting timing and understand each player's movement. When I find someone who fits the lineup, I prefer giving that card enough games to judge its real contribution. A few missed shots or one poor defensive possession are not necessarily reasons to replace an otherwise useful player.
    The market deserves patience as well. Before purchasing an upgrade, compare similar cards and think about how often you will use them. A high-profile player might be attractive, but a more affordable option could provide nearly everything your lineup needs. Leaving some resources unused is not a failure to make progress. It gives you the freedom to respond when a genuinely useful opportunity appears.
    For players who would rather spend more time playing than concentrating entirely on resource collection, purchasing additional currency may be worth researching. U4GM has a straightforward ordering process and competitive pricing, making it an option to consider when comparing available choices. Checking the latest product information and applicable game rules before proceeding is still important.
    I enjoy MyTEAM most when roster decisions lead to new ways of playing rather than simply increasing the overall rating displayed on the screen. A better defensive combination might encourage you to play more aggressively, while a reliable shooter could change how often you attack the basket. If you have a specific upgrade in mind and want to spend less time waiting to build the necessary resources, you can look into buy MT NBA 2K27 as one possible part of your roster plan.

  • 开源软件 | 开源社区 | 开源理念 | 开源与商业 | 开源可持续发展 等相关话的交流讨论
    注: 这里的"开源"是泛化的共建共享概念, 范围包含 OSI的范围、自由软件、CC等相关内容

    58 主题
    252 帖子
    VentusUtaV

    @sunrisepeak 在 如果你的开源软件被人分发,应该如何应对 中说:

    如果作者不愿意维护Safari版本. 在明确的告诉用户软件版权信息及改动, 以及符合基于版本的协议下, 有人fork维护Safari版本并进行收费也算正常

    主要是当时那人没遵守MIT协议保留原协议文本,且在App Store商店页把支持页面给写到了BewlyBewly的issue页面。

    本来调试Safari就烦,公然违反极为宽松的MIT算是压死骆驼最后一根稻草。

  • 67 主题
    278 帖子
    siconger005-beepS
    Hi there 👋 🚀 佳期投资正在寻找硬核技术极客!

    在招岗位:核心交易系统工程师(C++ / 低延迟 / Base 京沪)、AI Infra、存储工程师、SRE、MLE等。

    🏢 关于佳期 (About Jiaqi)

    佳期投资是一家百亿规模的量化对冲基金。十余年来,我们凭借全球领先的技术、卓越的数据研究能力,以及严谨、系统化的投资策略,持续创造优秀的投资业绩。我们的全明星团队汇聚了来自国内外顶尖学府的优秀人才,成员兼具丰富的量化投资经验与多元化的学科背景。在佳期,专业严谨、高效敏捷与团队协作并存,对技术的热爱与创新精神驱动着我们不断突破、持续前行。

    🔍 我们在寻找谁? (Who we are looking for) 💻 【核心交易系统工程师 (Core Trading System Engineer)】

    你的硬核战场 (Responsibilities):

    架构重塑:设计并搭建行业顶级的量化交易系统及底层基础设施架构。 极致调优:在不同编译器、操作系统(Linux Kernel)及网络环境下,进行极限性能压榨(如 CPU Cache 优化、内存屏障、NUMA 亲和性等)。 降维打击:追踪并引入其他高科技行业(如 HPC、底层数据库、游戏引擎)的尖端技术,跨越技术边界解决低延迟难题。

    我们期望的画像 (Requirements):

    基础扎实:国内外知名院校计算机相关专业,精通数据结构与算法,具备极其扎实的 C++ 编程能力(工程落地能力强,过往工作 C++ 占比高)。 底层狂热:具备低延迟/高并发系统开发经验,熟悉 Linux 高级特性(如 eBPF、内核旁路、用户态协议栈、DPDK)者优先。 极客精神:对技术创新有狂热追求,永远在寻找最高效、最优雅的底层解决方案。 🎁 你能得到什么? (Why join us?) 真正的技术挑战:拒绝 CRUD 和内卷,你的每一行底层优化都将直接转化为真金白银的收益。 顶配的团队:与来自顶尖学府及海内外大厂的极客一起工作,纯粹的工程师文化。 极具竞争力的回报:提供行业顶配的薪酬待遇与奖金机制,与你的技术产出直接挂钩。 📬 如何加入 / Contact

    如果你觉得你的技术栈与我们高度匹配,或者你是一个对底层性能有强迫症的硬核开发者,欢迎随时联系:

    Email: yyuan@jqinvestments.com (请附上简历,建议同时附上你的 GitHub 主页或技术博客) WeChat: 18861123700 (添加请备注:GitHub + 姓名 + 岗位)

    (注:佳期投资常年开放核心交易系统、AI Infra、存储、SRE、MLE 等技术岗位,欢迎随时来聊!)

  • 38 主题
    79 帖子
    DoomjustinD

    这是在学习了capy的隐式注入玩法后的更新。他的方案帮我梳理了我原本混乱的思路,感觉自己的目光都清撤了许多。

    旧版本的困局:stop_then 手动包装

    旧版本就是让用户手工拼装,每个 awaitable 都得用 stop_then 包一下:

    co_await stop_then(some_io_operation(), stop_token); // 得手动拼,十分繁琐

    这种操作确实实现了取消,但是所有的复杂性都暴露给了业务代码:

    编写 co_await 时需要考虑"这个操作是否需要取消?" 如果需要,必须记得手动包裹 stop_then。 遗漏任何一个,就会产生"无法取消的操作",取消时挂起。

    值得一提的是,原本的实现由于混乱的思路引入了各种多线程问题。但是在梳理之后发现都是不必要处理的操作,属于是自己把自己坑进去了。

    新版本的改进:隐式注入取消 消除业务层污染

    旧方案的致命缺陷不在技术复杂度,而在于污染。业务代码里到处是 stop_then,每个 I/O 函数签名都要添加 std::stop_token 参数,这直接增加了维护成本和开发者的认知负担。

    // 旧方案:取消把整个调用链都污染了 Task<> handle_client(IOContext& ctx, std::stop_token token) { auto res = co_await stop_then(async_read(fd, buffer), token); // 得手动拼 if (!res && res.error() == std::errc::operation_canceled) { // 处理被取消 } }

    我原本认为用户可能需要自己选择可取消和不可取消,所以通过stop_then包装器可以提供最大的自由度。但是实际上,我想不出什么操作是不需要被取消的,甚至更进一步的想,一旦出现了不可取消的挂起操作,某种意义上其实可以算作一种资源泄露bug了。

    想法来源:从 capy 的实现中得到启发,我发现可以用 C++20 的 await_transform 重新设计这一层。capy 为了通用性,通过 env 注入 executor、allocator、stop_token 等参数;而我们的框架强绑定 io_uring,所以能更直接地在 promise 层注入 IOContext 和 stop_token,并统一依赖 ASYNC_CANCEL 语义。正因为这些都能确定下来,所以取消机制才能被完全自动化。

    核心改进:三层隐式注入

    第一层:Promise 环境绑定

    每个 Task 的 promise 自动持有当前的 IOContext 和 stop_token,无需用户处理:

    struct StoppablePromise { IOContext& context_; std::stop_token stop_token_; // ... };

    第二层:await_transform 统一拦截

    利用 C++20 的 await_transform 机制,框架可以在业务代码的每一个 co_await IOAwaiter 处进行拦截,自动将其升级为"取消感知"版本:

    // 新模型:零污染,框架自动处理 Task<> handle_client() { auto res = co_await async_read(fd, buffer); // 框架自动注入取消机制 if (!res && res.error() == std::errc::operation_canceled) { // 处理取消情况 } }

    第三层:统一落地语义

    StopTokenWrapper 在 awaiter 挂起期间隐式注册取消回调。被取消操作的最终结果由原操作的 completion 决定:

    也是在重新设计这块时,我发现并不需要处理 ASYNC_CANCEL 的 cqe,直接丢掉,反而让整个实现简单起来了,没有复杂的生命周期问题,也没有了被取消操作和取消操作的时序问题,被 resume 时一定是被取消操作的结果返回了,而且被取消操作的结果也是精确地,result会被正确设置。

    template<typename Promise> auto await_suspend(std::coroutine_handle<Promise> handle) noexcept -> std::coroutine_handle<> { auto inner_handle = inner_.await_suspend(handle, context()); // 核心:在挂起期隐式注册取消回调 if (promise_->stop_token.stop_possible()) stop_callback_.emplace(promise_->stop_token, CancelFn{ this }); return inner_handle; } // 取消触发时的统一路径 void CancelFn::operator()() noexcept { if constexpr (cancellable<Awaitable>) wrapper->inner_.cancel(wrapper->context()); // 若 Awaiter 实现 cancel,直接调用 else wrapper->context().cancel(wrapper->inner_); // 否则走 ASYNC_CANCEL } 修改后

    隐式注入:用户编写普通的 co_await,框架在后台自动升级为"取消感知"版本。

    生命周期简化:由于 cancel CQE 直接扔掉,不再跟踪,stop_callback 的生命周期严格绑定在 awaiter 挂起这段期间。不需要 shared_ptr<bool> 守卫了。

    awaiter 契约统一:

    要是 awaiter 实现了 cancel(io_context) 方法,框架就调它。 否则用默认的 io_uring ASYNC_CANCEL 方式。 核心代码对比

    旧版本(用户手动拼):

    auto task = [](std::stop_token token) -> async::Task<> { co_await stop_then(some_io(), token); // 得手动传 token,手动调用 stop_then }; co_await async::any(task(group.stop_token()), ...);

    新版本(框架自动处理):

    auto task = []() -> async::Task<> { co_await some_io(); // 啥都不用干,框架已经全搞定 }; // stop_token 框架内部在 Task::promise 里已经绑了,用户看不见 多线程下的行为对比 旧版本 主线程 (IOContext.run()) ↓ [处理 I/O 完成事件A] ↓ task 协程被唤醒 ↓ stop_then 析构(alive_ = false) ↓ 后台定时器线程 ↓ [发送 stop_token cancel] ↓ post 任务进入队列 ↓ [IOContext 取出 post 任务,检查 alive_] ↓ *alive_ == false,安全丢弃

    陷阱:如果 post 任务执行前,协程没析构,就访问了已 free 的内存。

    新版本 主线程 (IOContext.run()) ↓ task 协程开始 co_await some_io() ↓ await_transform 自动把它包装成 StopTokenWrapper ↓ stop_callback 被注册在 wrapper 的栈帧上 ↓ 后台定时器线程 ↓ [stop_token 被触发] ↓ stop_callback 立刻在这个线程上执行(标准库保证线程安全) ↓ 调用 awaiter.cancel(context) 把 cancel 指令 post 回主线程 ↓ 主线程 (IOContext.run()) ↓ [从事件循环取出 cancel 指令] ↓ 发送 ASYNC_CANCEL SQE 给内核 ↓ 原操作的 CQE 返回 -ECANCELED 或原结果 ↓ awaiter 恢复协程,返回错误给上层

    如此一来:

    stop_callback 本身线程安全(标准库保证)。 cancel 动作在 IOContext 里执行,安全、可控。 栈帧生命周期自动保护,不需要 shared_ptr 守卫。 结尾

    由于这个修改算是撬了最底层的设计,所以导致绝大部分的模块都需要重写。不过在理清了混乱之后,所有的实现相较于旧版本都变得简单了很多,甚至绝大部分的Awaiter实现都变成了下面这种极简版本,也就是继承一下IOAwaiter,写个 prepare()和构造函数就完事了。

    class SendAwaiter : public async::IOAwaiter<SendAwaiter, std::size_t> { private: int fd_; std::span<const std::byte> buffer_; public: SendAwaiter(int fd, std::span<const std::byte> buffer) : fd_{ fd } , buffer_{ buffer } {} void prepare(::io_uring_sqe* sqe) const noexcept { ::io_uring_prep_send(sqe, fd_, buffer_.data(), buffer_.size(), MSG_NOSIGNAL); } };
  • 一个技术知识分享、学习、交流的社区

    16 主题
    54 帖子
    sunrisepeakS

    @dustchens 已支持

  • Got a question? Ask away!

    4 主题
    14 帖子
    SPeakS

    备注一下使数学公式的使用语法

    单行公式语法 - $ 你的公式 $

    $ log_2^n $

    $ log_2^n $

    多行公式语法 - $$ 你的公式 $$

    $$ log_2^n => log_2^9 = 3 , n = 9 $$

    $$
    log_2^n =>
    log_2^9 = 3, n = 9
    $$

公告栏 | Bulletin Board

欢迎加入d2learn社区 - 社区指南
Welcome to the d2learn Community - Community Guide

一个以 [知识、技术、代码、项目、想法、开源] 相关话题为主导的社区
A community focused on topics related to [knowledge, technology, code, projects, ideas, and open source].


在线用户