今日重点:🗣️ 口语——Sasha Levin 10-25 发出的 pull request,11-17 只在邮件线程里补了一句 Ping?。1 小时 22 分钟后 Linus 回了第一封,又过 5 分钟补了第二封,把「内核 API 规范框架」 整个否决。两封邮件没有任何寒暄,第一句就是结论,然后逐层钉死「什么不行、什么才算行」。 这是「怎么在评审里说不」的完整口语范本——今天拆六个动作,每个都能搬到你的下一次评审会。 三封邮件全部逐字摘自 lore 原文(11-17 02:14 / 03:36 / 03:41 北京)。
🗣️ 现场:一句 Ping? 与两封否决
/ section背景:Sasha Levin 在 10-25 为 v7.4 发出 [GIT PULL] Kernel API specification framework—— 把 API 规范写成源码里的结构化 kerneldoc,构建时转成 vmlinux 里的数据,再通过 debugfs 暴露给用户态。 11-17 02:14(北京),他在线程里补了一个词加一个问号:Ping?。 1 小时 22 分钟后,Linus 在 03:36 回了第一封,03:41 又补了第二封。先看第一封正文全文——注意它没有「感谢提交」之类的客套, 第一句就是结论。
So honestly, I really don't like it. I don't think *any* of this should be inside the kernel. I don't think the in-kernel side helps at all: the comments are too big, nobody will read them, and the whole - and only - real use of them would be outside validation, not at run-time.
IOW, I see only downsides to polluting the kernel source code with comments that aren't useful to humans.
If this was a spec that is maintained in some automated format - not in C comments in source files that aren't useful as comments anyway purely inside something like tools/kernel-api/ (or Documentation, or even an outside repository entirely), together with the scripts to then validate it, that would be one thing.
But polluting source code with huge comments that aren't for humans is right out. Particularly when it clearly is a "first step" and the intent is to make those comments bigger and more extended in the future.
And I see *no* upside to any /sys/kernel/debug/kapi interface. Nobody cares, and any excuse to make it dynamic ("We might update the API in the future") is actually the complete opposite of the whole point of a stable API and not something I want to make easier or support in any way what-so-ever.
So all of the in-kernel side to me looks like actively detrimental to the whole idea.
What I think *can* work is an external spec that is written so that tools (read: AI) can actually then verify it against kernel sources, and that can be used to also drive automated testing. But putting any of it inside the kernel is bad.
Linus
🗣️ 六个动作
/ section① 定调:So honestly, I really don't like it.
语境翻译:说实话,我真心不喜欢这个东西。
要点:三个词各干一件事——So 是口语里的收线词,把前面的对话收起来,表示「我要说正事了」; honestly 提前声明「下面不绕弯」;really 把 dislike 加重。整句没有一个术语,却是全场最重的表态。 语序更值得学:结论放在最前,理由在后面(下一段才讲 comments 太大、没人读)。口语里这叫「先落锤、再举证」—— 评审会上最常见的错误是先铺垫三分钟背景,听的人已经不知道你要说什么了。
复述引导:用同样的语序说一句你自己的否决——「So honestly, I really don't like [具体东西].」 重音落在 really 和 don't 上,So 后面停半拍。
来源:Linus Torvalds,11-17 03:36 北京
② 钉死范围:I don't think *any* of this should be inside the kernel.
语境翻译:我认为这里头没有任何一样东西该进内核。
要点:邮件里用星号包住 *any*,等于说话时重读 any——邮件没有声调,就用星号代替重音。 这一个重音把否定的范围从「有些地方要改」扩到「整个方向都不该在这里」。紧接着三条理由全是口语短句: the comments are too big、nobody will read them、 the whole - and only - real use of them would be outside validation, not at run-time。 三个短句连排,中间用 - and only - 插入强调——写出来像停顿,念出来就是放慢语速、一字一顿。
复述引导:说「I don't think any of this should be [你认为不该在的位置].」,后面接三个短理由, 每个不超过八个词。练的是短句连排的节奏,不是词汇量。
来源:同第一封
③ 一句话收口:IOW, I see only downsides to ...
语境翻译:换句话说,用对人类没用的注释去污染内核源码,我只看到坏处。
要点:IOW = in other words,邮件里的口语缩写,念出来就是「换句话说」。这是评审发言里最实用的动作: 讲完几条理由,用一句 IOW 把论证压成一句可复述的话,对方没法装听不懂。这句最重的词是 polluting——不是 adding,不是 putting;选了 polluting 就等于给这件事定了性。 但后面马上用 comments that aren't useful to humans 把定性限定在具体对象上:否定的是做法,不是人。
复述引导:先给两条理由,再说「IOW, I see only downsides to [动名词短语].」—— IOW 后面接动名词结构,这是把动作名词化的英语习惯。
来源:同第一封
④ 划边界再关门:that would be one thing ... is right out.
原句:If this was a spec that is maintained in some automated format ... together with the scripts to then validate it, that would be one thing. But polluting source code with huge comments that aren't for humans is right out.
语境翻译:如果这是一份用自动化格式维护的规范……再配上校验脚本,那倒另说。但拿大段不是给人看的注释污染源码,绝对不行。
要点:整封邮件里最值得背的句式。that would be one thing = 「那倒另说」, 先给对方留一条出路,表明「我不是否定你的目标」;is right out = 英式口语「绝对不行、直接出局」, 比 unacceptable 更口语,也把讨论空间关得更死。两个动作连起来:先划出可接受边界 (外部仓库 / tools/kernel-api / Documentation + 校验脚本),再把当前形式关在门外。 否决提案最忌只说「不行」;这句给的是「这样不行,那样可以」。
复述引导:说「If this was [可接受的形式], that would be one thing. But [当前形式] is right out.」—— 中间必须具体到可执行的形式,不能只说「换个方式」。
来源:同第一封
⑤ 短句连击:Nobody cares. / Completely pointless.
原句:And I see *no* upside to any /sys/kernel/debug/kapi interface. Nobody cares, and any excuse to make it dynamic ("We might update the API in the future") is actually the complete opposite of the whole point of a stable API and not something I want to make easier or support in any way what-so-ever.
语境翻译:那个 /sys/kernel/debug/kapi 接口我看不到任何好处。没人在乎。而且任何把它做成动态接口的借口(「我们以后可能会更新 API」),恰恰和稳定 API 的初衷完全相反——我不但不想让它更容易,任何形式的支持都不想要。
要点:Nobody cares. 三个词独立成句。口语里短句不代表话少,是收掉音量反而更重—— 前面刚说完长句,突然三个词一顿,听众会本能地重视。后面 in any way what-so-ever 把 what-so-ever 拆开加连字符,念的时候一字一顿。注意他质疑的不是实现,而是需求本身: 「没人需要它」比「它写得不好」更难反驳。
复述引导:两句连说——「I see no upside to [东西]. Nobody cares.」第二句刻意放慢、降调。
来源:同第一封
⑥ 对仗收尾:The spec needs to match reality, we're not changing reality to match a spec.
原句(第二封):The kernel API has grown over time, and pretending anything else would make the whole project be just a fever dream. We are *NEVER* going to change the API _to_ some spec. The spec needs to match reality, we're not changing reality to match a spec. ... But neither of those phases are ever going to involve any humans, because competent humans have better things to do.
语境翻译:内核 API 是一路长出来的,假装不是这样,整个项目就成了高烧时的梦话。我们永远不会为了让 API 符合某份规范去改 API。规范要对齐现实,我们不会改现实去对齐规范。……而且这两个阶段都不会涉及任何人,因为能干的人有更要紧的事做。
要点:三个口语技巧叠在一起。fever dream(高烧时的梦)是形象化比喻,比 unrealistic 更有画面; *NEVER* 全大写加星号 = 说话时把这个词砸出来;最后一句是英语口语里最好用的修辞—— 正着说一遍,再反着说一遍(spec matches reality / not changing reality to match spec), 结构对称,听众一次就记住。收尾的 competent humans have better things to do 是幽默式否定: 不直说「人工检查不现实」,而说「能干的人有更要紧的事做」。
复述引导:把对仗句念三遍,直到能脱口而出——「X needs to match Y, we're not changing Y to match X.」 再把 X/Y 换成你的场景,例如「The driver needs to match the DT binding, we're not changing the DT binding to match the driver.」
来源:Linus Torvalds,11-17 03:41 北京
为什么这六个动作值得背。否决不是把话说重,而是把话说准:先给结论(①),把范围钉死(②), 一句话收口(③),划出唯一可接受的形式(④),直接质疑需求本身(⑤),最后用对仗句给整件事定调(⑥)。 全程没有一句「I think maybe」——口语化的英语不等于含糊的英语。
✨ 辅助彩蛋:四个能直接搬走的词
/ column★ 四个词,各管一种语气
🔗 原文
✍️ 今日练习
/ column💡 显示答案
每日一句:Say no in three moves — name it, bound it, leave one door open.
素材:lore.kernel.org · linux-mm 11-17 窗口(Sasha Levin 的 Ping? · Linus 两封否决)· 内核英语 · 每日一篇