全部文章
-
Leaderboard walk-through: score events, a derived sorted set and rebuildable rankings
A design walk-through for leaderboards: server-authoritative score events with an idempotency key as the truth, one sorted set per board period as a derived index answering top-N and single-member rank queries, tie rules encoded into the score, period boundaries routed by occurrence time, and rebuilds as a routine.
-
Redirects 301, 302, 307 and 308: which ones preserve the request method
301 and 302 permit a user agent to turn a redirected POST into a GET; 307 and 308 forbid changing the method, so the body is resent; 303 always switches to a GET of another resource. 301 and 308 are heuristically cacheable, 302 and 307 are not. Choose by permanence and by whether the method must survive.
-
在入口处一次性配置 logging 模块
库只获取日志记录器，不配置处理器；应用在启动时统一配置一次处理器、级别和格式，使输出集中受控，并避免重复日志。
-
Reading the load average and understanding the OOM killer
Linux load average counts runnable and uninterruptible-sleep tasks over 1, 5 and 15 minutes, so a high number with idle CPUs points at I/O or a hung filesystem. When memory cannot be reclaimed, the OOM killer picks a task by badness score, which oom_score_adj shifts from -1000 (never) to +1000 (first).
-
回答新人最初五个问题的 CONTRIBUTING 文件
准备贡献的人在写代码前会问：项目需要这个变更吗，如何提出，拉取请求必须包含什么，多久有人回复，合并后的变更如何到达用户。CONTRIBUTING 文件按顺序回答这五个问题，其余内容通过链接提供，旨在预先避免因超出范围或缺少测试而被拒绝的拉取请求。
-
Stereoscopic XR Rendering: Preserve Per-View Poses, Projections and Image Layers
Render consistent stereo views using runtime view data and validate multiview optimizations without losing eye-specific geometry or effects.
-
将缺陷报告转化为回归测试
修复缺陷前，先用一个注明报告的失败自动化测试复现它；这个测试既证明修复有效，也防止缺陷再次出现。
-
LoRA：低秩适配及其内存节省的真实范围
LoRA 冻结基础权重，学习低秩更新。它减少可训练参数及其优化器状态，但基础模型、激活和部署兼容性仍然需要考虑。
-
Unreal gameplay framework: place rules, shared state and possession deliberately
Separate server rules, replicated match information, player control and the possessed body so respawning and multiplayer do not mix their responsibilities.
-
What share of shipped features are still used 90 days after release, and what happened to the unused ones and their flags?
Open question: feature-flag tools such as Unleash attach an expected lifetime to a flag and mark it potentially stale once that lifetime passes, which covers the flag but not the feature; for products that track feature usage, what share of features released in a year had meaningful use after 90 days, what happened to the rest (removed, hidden, kept), and did their flags and code follow?
-
使用 debsums 和 dpkg --verify 检查 Debian、Ubuntu 已安装包的完整性
debsums 将已安装文件与软件包构建时记录的 MD5 校验和进行比较；dpkg --verify（自 dpkg 1.17.2 起）比较 dpkg 数据库本身记录的元数据；dpkg -S/-L 则在决定是否重新安装之前回答“哪个包拥有这个文件”和“这个包安装了哪些文件”。
-
Managing certificates from PowerShell: the Cert: drive, PFX import/export, and binding one to IIS
Browsing Cert:\LocalMachine\My, filtering by expiry with Get-ChildItem, moving a certificate with its private key via Import-PfxCertificate/Export-PfxCertificate, New-SelfSignedCertificate for test use only, and binding the result in IIS with New-WebBinding or checking it with netsh http show sslcert.
-
State units beside every measurement
Make measurement records interpretable by keeping units, aggregation and population beside each value.
-
IBM i security basics for administration: special authorities, object authority, QSECURITY, and QAUDJRN
IBM i access control rests on special authorities such as *ALLOBJ and *SECADM granted per user profile, object-level authorities checked with DSPOBJAUT/EDTOBJAUT, the QSECURITY system value that sets the enforcement level, and the QAUDJRN audit journal that records what happened — all administration and auditing, no exploitation.
-
Logging a consumer PM2.5 sensor: what to record beside each reading and what the number cannot say
A consumer particulate sensor reports an estimate of the mass of particles 2.5 micrometres and smaller; the EPA states that such sensors come with lower accuracy and higher uncertainty than regulatory monitors, describes collocation with reference instruments as an effective way to assess them, and notes that an outdoor correction may not apply indoors and that changing temperature and humidity may affect some sensors. A household log therefore stores the raw value, the averaging interval, placement, indoor events and the nearest official monitor's value with its distance.
-
Testing authorization through resource relationships rather than role names
Check whether a caller can act on a particular object through the relationship the product actually promises. This proposed lab method treats role labels as fixture attributes, not as the test oracle.
-
Modelling states with Literal, Enum and TypedDict
Replace free strings and loose dictionaries with Literal unions, enumerations and TypedDicts so that impossible states are rejected by the type checker and documented in one place.
-
make as a task runner: phony targets, tabs and one shell per line
A Makefile is a workable entry point for a repository's commands if command targets are declared .PHONY, recipe lines start with a tab, and it is understood that each recipe line runs in its own shell unless .ONESHELL is set; keep the default goal harmless and keep real file dependencies real.
-
Foot placement IK: preserve contact without fighting the walk cycle
Build terrain-aware foot targets, respect leg reach and contact phases, and keep pelvis adjustment and moving-platform anchors consistent.
-
Recursion versus iteration: stack depth, limits and when to convert
Recursion mirrors tree-shaped problems but spends one stack frame per level; the stack is bounded by the interpreter's recursion limit or the thread's stack size, so recursion over input-controlled depth is a crash waiting to happen. Convert to an explicit stack or loop when depth grows with input size.
机器可读： JSON