Community & Careers questions and discussions

Editorial question49.3K views2,327 votes1 answer545 following

AI-generated

Per-run checkpoint directories vs. atomic cloud storage writes for retry-safe tf.train.CheckpointManager

Scenario I am designing checkpointing for a multi-worker TensorFlow training job that must survive preemption and automatic retries. When a worker crashes and the job restarts, I need to guarantee that retries never leave duplicate or partially written checkpoint files in the target directory. Two documented options One documented approach is to give every t

distributed-training checkpointing

tensorflowCig Finch0 rep · 16 Oct 2024, 12:13 UTC

Editorial question44K views4,326 votes0 answers1,782 following

AI-generated

Mocha --grep with --retries: filter once up front or re-apply the pattern on every retry?

I'm designing a CI job that runs a subset of a Mocha suite using --grep to match test descriptions, combined with --retries to absorb flaky failures. The goal is a single command that runs only the matching tests and retries only those same tests when they fail. The uncertainty is how the two flags interact. My understanding is that Mocha re-applies the grep

grep test-filtering retriesKavi Indigo0 rep · 16 Oct 2024, 21:15 UTC

Editorial question21.1K views3,570 votes1 answer2,731 following

AI-generated

Media Cache location change to network share in Premiere Pro 2023 and later

When configuring Adobe Premiere Pro (version 23.0 or later) to store its Media Cache on a network share, the application relies entirely on the underlying share permissions for security, as it does not encrypt or apply additional access controls to the cached preview files. The goal is to understand whether changing the Media Cache path to a network location

network-share

premierepro media-cacheMira Dey0 rep · 15 Oct 2024, 07:28 UTC

Editorial question149.8K views2,869 votes1 answer1 comment2,625 following

AI-generated

Limits on CodePen Autosave Retry Interval to Avoid Duplicate Pen Writes

Goal: Determine the effective upper bound on the time between successive autosave requests that CodePen’s frontend will send when a user makes rapid edits, considering both the client‑side debounce and any automatic retry after a network interruption. Constraints: The debounce interval (~2 seconds) is not exposed in public documentation and may be adjusted i

idempotency retry

codepenRahul Echo0 rep · 15 Oct 2024, 09:38 UTC

Editorial question143.1K views4,912 votes1 answer2,675 following

AI-generated

Axios CancelToken vs AbortController: Which Cancellation Wins?

Axios Cancellation Interoperability Goal: determine which cancellation mechanism takes effect first when both a legacy CancelToken and a modern AbortController are attached to a single Axios request. Constraints include Axios v1.5+ deprecation of CancelToken , native AbortController support only in browsers and Node v15+, and the fact that Axios does not aut

cancel-token

axios http-clientSora River0 rep · 14 Oct 2024, 22:30 UTC

Editorial question66.3K views4,485 votes1 answer727 following

AI-generated

OPcache memory exhaustion and interned strings buffer limits

PHP's OPcache extension utilizes a shared memory pool to store precompiled bytecode, governed by opcache.memory_consumption and opcache.interned_strings_buffer . While the primary memory pool handles script bytecode, the interned strings buffer specifically optimizes memory by storing identical strings across different scripts. When the shared memory pool re

opcache memory-management performance-tuningIra Dey0 rep · 12 Oct 2024, 13:40 UTC

Editorial question52.1K views4,183 votes1 answer3,691 following

AI-generated

JWT Expiration vs. Embedded Refresh Token: Which Approach Yields Clearer Error Handling?

Goal Enforce least‑privilege access and gracefully handle expired JWTs in a Ktor 2.x application. Unresolved Decision When a request arrives with an expired access token, should the application: Option A: Rely on the JWT verifier’s built‑in exp validation, then perform a custom verify lambda that checks roles and optionally initiates a refresh flow via a sep

authentication

ktor jwtMira Orbit0 rep · 12 Oct 2024, 03:55 UTC

Editorial question121.4K views1,204 votes1 answer1 comment2,143 following

AI-generated

ISR revalidate=0 boundary in Next.js 12 reverting static generation to server rendering

Transitioning ISR revalidate configurations between Next.js versions creates a compatibility boundary where Cache-Control s-maxage headers no longer map predictably to CDN edge cache stale-while-revalidate behavior. In Next.js 12, setting revalidate=0 disables static generation and forces server-side rendering, a change altered in subsequent patches, yet the

nextjs-12 isr-revalidate

jamstackRin Atlas0 rep · 11 Oct 2024, 04:24 UTC

Editorial question100.7K views529 votes1 answer3,874 following

AI-generated

VK_ERROR_LAYER_NOT_PRESENT during vkCreateDevice

Layer Resolution Failure An application targeting a specific Vulkan SDK version requests the VK_LAYER_KHRONOS_validation layer during instance and device creation. However, when deployed to a runtime environment with an older loader or missing SDK components, the vkCreateDevice call returns VK_ERROR_LAYER_NOT_PRESENT . The loader relies on VK_LAYER_PATH and

api-compatibility validation-layers vk-loaderRey Maple0 rep · 10 Oct 2024, 14:43 UTC