ReadMeFeed / Community knowledge

Questions & conversations

Real questions. Useful conversations. Find the people who know your stack.

Editorial question16.8K views4,255 votes1 answer3,027 following

AI-generated

Bun's --hot Reloader and the ESM Module Registry: Which Imports Get Invalidated?

Bun exposes two overlapping mechanisms for loading application code: its transpiling module loader, which caches compiled output so repeat imports are faster, and its watch/hot reload modes ( bun --watch , bun --hot ), which re-execute or patch modules when files on disk change. The boundary between these two mechanisms is not fully specified. In Bun 1.x, ho

esm module-cache hot-reloadCig Vale0 rep · 08 Jan 2026, 14:39 UTC

Editorial question107.6K views3,701 votes1 answer2,814 following

AI-generated

URLProtocol stubbing limits for background sessions in Swift

Testing without Production Credentials Swift developers often need to validate network logic without hitting live services. A common technique is subclassing URLProtocol to intercept URLRequest objects and return canned responses. The stub is registered in the test target’s URLSessionConfiguration , leaving production code untouched. Known Constraint While t

network testing

swiftDev Byte0 rep · 08 Jan 2026, 14:48 UTC

Editorial question79.1K views2,024 votes1 answer3,183 following

AI-generated

Does norg provide a stable parser API for reproducible document processing?

Goal: Ensure that a development team can share norg files and obtain identical parsed representations regardless of the host machine or the version of the norg parser installed. Constraint: The norg parser is actively evolving, and recent releases have adjusted how certain constructs—such as nested heading blocks, custom macros, and link resolution—are trans

ast reproducibility

norgSky Indigo0 rep · 07 Jan 2026, 15:06 UTC

Editorial question82.5K views948 votes0 answers1 comment3,318 following

AI-generated

Silent vCPU Overcommit: No Warning When %RDY Latency Exceeds Threshold

VMware Fusion exposes the CPU ready metric (%RDY) in performance charts and optionally in vmware.log when sched.stats.enable = TRUE is set. Under concurrent multi-VM workloads, %RDY rises once combined vCPU demand exceeds physical cores, yet Fusion emits no dedicated log entry or alert when this latency crosses a problematic threshold. The hypervisor permits

cpu-ready scheduling-latency vmware-fusionLax Sage0 rep · 07 Jan 2026, 18:21 UTC

Editorial question78.2K views3,911 votes0 answers1,946 following

AI-generated

process.stdout.write vs console.log for crash log persistence

Choosing between synchronous process.stdout.write and asynchronous console.log for crash diagnostics In containerized deployments the output stream is usually a pipe, so console.log buffers writes asynchronously. If the process crashes — for example from an OOM kill or an unhandled exception — the final stack trace can stay in the buffer and never reach the

nodejs logging

javascriptRey Nova0 rep · 07 Jan 2026, 15:45 UTC

Editorial question148.5K views1,275 votes1 answer677 following

AI-generated

Ensuring data integrity verification after Shopware 6 point-in-time restore

Goal Establish a reliable method to verify that a restored Shopware 6 database and file system from a point-in-time backup contain complete and untampered data. Constraints Shopware 6 core provides no built-in backup/restore module; recovery relies on manual database dumps and file copies. Mixing backups from different timestamps causes media references to m

restore data-integrity

shopwareZo Finch0 rep · 07 Jan 2026, 06:15 UTC

Editorial question129.9K views1,912 votes0 answers1,371 following

AI-generated

XMI Export from Papyrus to Enterprise Architect: Do stereotypes and diagram layout survive round‑trip?

Goal Validate that a UML model exported as XMI from Eclipse Papyrus can be imported into Sparx Enterprise Architect without losing applied stereotypes, tagged values, or diagram layout information. Constraints & Uncertainty The XMI header specifies the UML metamodel version; a mismatch between Papyrus (e.g., UML 2.5) and Enterprise Architect may cause el

xmi modeling-tools umlZo Byte0 rep · 06 Jan 2026, 20:45 UTC

Editorial question128K views710 votes1 answer4,194 following

AI-generated

C# 8.0 NRT boundary handling for legacy libraries without nullable annotations

The Challenge When enabling nullable reference types on an existing codebase, the compiler treats unannotated legacy libraries as 'oblivious'—assuming T? for all reference types. This suppresses warnings but erases static safety guarantees at the call boundary between annotated and unannotated code. The four-state nullable context model (annotation context a

csharp nrt compiler-warningsRahul River0 rep · 05 Jan 2026, 22:00 UTC

Editorial question128.3K views3,553 votes1 answer2,068 following

AI-generated

DQL JOIN FETCH or fetch=EAGER mapping for a collection-heavy entity that is sometimes only counted

Scenario An application built on Doctrine ORM has an entity with a large one-to-many collection. Some code paths iterate the full collection, while others only need its size. The default lazy loading triggers one extra query per access, and in detached states it risks a LazyInitializationException, so a more deliberate fetching strategy is needed. The trade-

doctrine-orm join-fetch fetch-modeLax River0 rep · 05 Jan 2026, 15:55 UTC