Glossary: Rust, LLM, App Store and Steam terms

A

• App ReviewApp Review is Apple's check of every app version, in-app purchase and event before it reaches the App Store.
• App Review BoardThe App Review Board hears appeals of App Review rejections.
• App Store ConnectApp Store Connect is Apple's website and API for publishing apps: app records, builds, metadata, screenshots, prices, in-app purchases, TestFlight and review submissions.
• App Store Connect API key → full pageThe .p8 key, key ID and issuer ID that let scripts and CI publish apps.
• App Store review time → full pageHow long App Review takes, from real decision emails.
• App Store Small Business ProgramDevelopers who earned up to $1 million in App Store proceeds in the previous calendar year can enroll to pay a 15% commission instead of 30% on paid apps and in-app purch

B

• Batch APIA batch API takes many requests in one file and returns results later (within 24 hours) at a discount, typically half the normal token price..

C

• cargo publish → full pageUploading a crate to crates.io, step by step.
• Cargo.tomlCargo.toml is a Rust package's manifest: name, version, edition, rust-version, dependencies and features, written in TOML..
• CI (continuous integration)CI is a server that builds and tests your code on every push.
• Coding agent (agent step)A coding agent is an LLM in a loop that reads your repo, edits files and runs commands until a task is done.
• Context windowThe context window is the most tokens a model can take in one request, prompt plus answer.
• Crate (Rust)A crate is the unit Rust compiles and shares: one library or one binary, with its own Cargo.toml.
• crates.iocrates.io is the Rust community's package registry: cargo downloads dependencies from it and cargo publish uploads to it.
• crates.io download count → full pageWhy the counter overstates real use, and by how much.
• Cron expressionA cron expression is five fields, minute hour day-of-month month day-of-week, that say when a scheduled job runs.

D

• DLC (downloadable content)DLC is extra content sold for a game you already have: expansions, cosmetics, soundtracks or supporter packs.
• docs.rsdocs.rs builds and hosts API documentation for every crate published to crates.io, for every version, automatically..

E

• EmbeddingsAn embedding is a list of numbers that represents the meaning of a text, so similar texts get nearby vectors.
• Exponential backoff → full pageWaiting longer after each failed retry, plus jitter so clients do not retry in sync.

F

• fastlanefastlane is an open-source tool that automates iOS and Android release chores: signing, building, screenshots, metadata upload and submitting for review..

G

• Guideline 2.1 (information needed) → full pageApp completeness, and the questions every new account gets.
• Guideline 4.3 (spam) → full pageApple's rule against duplicate and look-alike apps.

H

• HTTP 429 Too Many Requests (rate limit)HTTP 429 means you sent too many requests in a time window.

I

• In-app purchase (IAP)An in-app purchase is digital content or features sold inside an iOS app through Apple's payment system: consumables, non-consumables, auto-renewable and non-renewing sub
• Inference (AI)Inference is running a trained model to get an output, as opposed to training it.

J

• Jitter → full pageRandomness added to retry delays so many clients do not retry at the same moment.
• JWT (JSON Web Token)A JWT is a signed token in three base64url parts, header.payload.signature, defined in RFC 7519.

L

• LiteLLM → full pageThe LLM gateway library, and its supply chain incident.
• Live ActivitiesLive Activities show an app's ongoing state (a timer, a ride, a score) on the Lock Screen and in the Dynamic Island, updated by the app or by push.
• LLM API Price Index → full pageOur monthly index of what 1,000 LLM requests cost.
• Long-context pricingLong-context pricing is a higher per-token rate some models charge when a single prompt is longer than a threshold, for example over 100,000 or 200,000 input tokens.

M

• MCP (Model Context Protocol)MCP is an open protocol, introduced by Anthropic in November 2024, for connecting AI apps to tools and data: a server exposes tools and resources, and any MCP client (Cla
• MSRV (minimum supported Rust version)MSRV is the oldest Rust compiler a crate promises to build on.

N

• Non-consumable (one-time unlock)A non-consumable in-app purchase is bought once and kept forever, across the user's devices: a Pro unlock, an ad remover, a level pack.

O

• Open weightsAn open-weights model publishes its trained weights so anyone can download and run it (Llama, DeepSeek, Mistral, Qwen).

P

• Prompt caching (cache read)Prompt caching lets a provider reuse the processed start of a prompt you send again (a system prompt, a document, an agent's history).

Q

• Quiet-day baseline → full pageMedian daily downloads on days with no release nearby: our steady-use number for a crate.

R

• RAG (retrieval-augmented generation)RAG answers a question by first retrieving relevant passages from your own documents (usually by embedding search) and putting them in the prompt, so the model answers fr
• Release window (crate downloads)Our term: the day a crate version is published plus the next day.
• Reverse dependenciesA crate's reverse dependencies are the other published crates that list it as a dependency.
• Rust editionAn edition (2015, 2018, 2021, 2024) is an opt-in set of language changes set per crate with edition = "2021".

S

• Semantic cache → full pageReusing an LLM answer for a prompt that means the same as one already answered.
• Semantic versioning (semver)Semantic versioning numbers releases MAJOR.MINOR.PATCH: breaking changes bump MAJOR, new compatible features bump MINOR, fixes bump PATCH.
• Snowflake IDA Snowflake ID is a 64-bit ID that sorts by time: a 41-bit millisecond timestamp, a 10-bit machine ID and a 12-bit sequence.
• Steam capsule imageCapsules are the store art Steam shows in lists, search, the front page and the library, each at a fixed pixel size (for example the 462x174 small capsule and the 920x430
• Steam Next FestSteam Next Fest is a week-long event, held several times a year, where unreleased games offer free demos and run livestreams, featured across the Steam store..
• Steam visibility rounds → full pageSteam's rounds of free store exposure for a new game.
• Steam wishlistA Steam wishlist is a player saving a game to buy later.
• SteamDBSteamDB is an independent database that tracks every Steam app's updates, prices, player counts and store changes from public Steam data.
• SteamworksSteamworks is Valve's developer side of Steam: the partner site for store pages, builds, pricing and events, and the SDK for achievements, cloud saves and multiplayer..
• StoreKit 2StoreKit 2 is Apple's Swift API for in-app purchases and subscriptions: load products, buy, verify transactions on device, and check entitlements with async/await..
• SwiftUISwiftUI is Apple's declarative UI framework: you describe views as functions of state and it updates the screen when the state changes.
• System promptThe system prompt is the standing instruction sent at the start of every request: the role, rules, tone and tools.

T

• TestFlightTestFlight is Apple's beta channel: internal testers (your team) get builds right away, external testers (up to 10,000) after a short beta review..
• Token (LLM)A token is the chunk of text an LLM reads and writes, roughly 3 to 4 characters of English on average.
• TOMLTOML (Tom's Obvious Minimal Language) is a config file format of keys, values and [tables].

U

• Unix timestampA Unix timestamp counts seconds since 00:00:00 UTC on January 1, 1970, ignoring leap seconds.
• UUIDv4 → full pageThe random UUID; why databases prefer v7 for primary keys.
• UUIDv7 → full pageA time-ordered UUID; how it differs from random UUIDv4.

V

• VRAM (for running LLMs)VRAM is the memory on a graphics card.

W

• WidgetKitWidgetKit is Apple's framework for Home Screen and Lock Screen widgets: small SwiftUI views refreshed on a timeline the system controls..

X

• x402 → full pageAn open standard that revives HTTP 402 Payment Required so APIs and agents can pay per request.
• XcodeGenXcodeGen generates an Xcode project file from a short project.yml, so the project is readable text instead of a hand-edited .xcodeproj..

Y

• Yanked version (cargo yank)Yanking marks a published crate version so new projects will not pick it, without deleting it: existing Cargo.lock files that already use it keep working..

By topic

Rust and crates

• cargo publish → full pageUploading a crate to crates.io, step by step.
• Cargo.tomlCargo.toml is a Rust package's manifest: name, version, edition, rust-version, dependencies and features, written in TOML..
• Crate (Rust)A crate is the unit Rust compiles and shares: one library or one binary, with its own Cargo.toml.
• crates.iocrates.io is the Rust community's package registry: cargo downloads dependencies from it and cargo publish uploads to it.
• crates.io download count → full pageWhy the counter overstates real use, and by how much.
• docs.rsdocs.rs builds and hosts API documentation for every crate published to crates.io, for every version, automatically..
• MSRV (minimum supported Rust version)MSRV is the oldest Rust compiler a crate promises to build on.
• Quiet-day baseline → full pageMedian daily downloads on days with no release nearby: our steady-use number for a crate.
• Release window (crate downloads)Our term: the day a crate version is published plus the next day.
• Reverse dependenciesA crate's reverse dependencies are the other published crates that list it as a dependency.
• Rust editionAn edition (2015, 2018, 2021, 2024) is an opt-in set of language changes set per crate with edition = "2021".
• Semantic versioning (semver)Semantic versioning numbers releases MAJOR.MINOR.PATCH: breaking changes bump MAJOR, new compatible features bump MINOR, fixes bump PATCH.
• Yanked version (cargo yank)Yanking marks a published crate version so new projects will not pick it, without deleting it: existing Cargo.lock files that already use it keep working..

LLM APIs and AI

• Batch APIA batch API takes many requests in one file and returns results later (within 24 hours) at a discount, typically half the normal token price..
• Coding agent (agent step)A coding agent is an LLM in a loop that reads your repo, edits files and runs commands until a task is done.
• Context windowThe context window is the most tokens a model can take in one request, prompt plus answer.
• EmbeddingsAn embedding is a list of numbers that represents the meaning of a text, so similar texts get nearby vectors.
• Inference (AI)Inference is running a trained model to get an output, as opposed to training it.
• LiteLLM → full pageThe LLM gateway library, and its supply chain incident.
• LLM API Price Index → full pageOur monthly index of what 1,000 LLM requests cost.
• Long-context pricingLong-context pricing is a higher per-token rate some models charge when a single prompt is longer than a threshold, for example over 100,000 or 200,000 input tokens.
• MCP (Model Context Protocol)MCP is an open protocol, introduced by Anthropic in November 2024, for connecting AI apps to tools and data: a server exposes tools and resources, and any MCP client (Cla
• Open weightsAn open-weights model publishes its trained weights so anyone can download and run it (Llama, DeepSeek, Mistral, Qwen).
• Prompt caching (cache read)Prompt caching lets a provider reuse the processed start of a prompt you send again (a system prompt, a document, an agent's history).
• RAG (retrieval-augmented generation)RAG answers a question by first retrieving relevant passages from your own documents (usually by embedding search) and putting them in the prompt, so the model answers fr
• Semantic cache → full pageReusing an LLM answer for a prompt that means the same as one already answered.
• System promptThe system prompt is the standing instruction sent at the start of every request: the role, rules, tone and tools.
• Token (LLM)A token is the chunk of text an LLM reads and writes, roughly 3 to 4 characters of English on average.
• VRAM (for running LLMs)VRAM is the memory on a graphics card.

App Store and iOS

• App ReviewApp Review is Apple's check of every app version, in-app purchase and event before it reaches the App Store.
• App Review BoardThe App Review Board hears appeals of App Review rejections.
• App Store ConnectApp Store Connect is Apple's website and API for publishing apps: app records, builds, metadata, screenshots, prices, in-app purchases, TestFlight and review submissions.
• App Store Connect API key → full pageThe .p8 key, key ID and issuer ID that let scripts and CI publish apps.
• App Store review time → full pageHow long App Review takes, from real decision emails.
• App Store Small Business ProgramDevelopers who earned up to $1 million in App Store proceeds in the previous calendar year can enroll to pay a 15% commission instead of 30% on paid apps and in-app purch
• fastlanefastlane is an open-source tool that automates iOS and Android release chores: signing, building, screenshots, metadata upload and submitting for review..
• Guideline 2.1 (information needed) → full pageApp completeness, and the questions every new account gets.
• Guideline 4.3 (spam) → full pageApple's rule against duplicate and look-alike apps.
• In-app purchase (IAP)An in-app purchase is digital content or features sold inside an iOS app through Apple's payment system: consumables, non-consumables, auto-renewable and non-renewing sub
• Live ActivitiesLive Activities show an app's ongoing state (a timer, a ride, a score) on the Lock Screen and in the Dynamic Island, updated by the app or by push.
• Non-consumable (one-time unlock)A non-consumable in-app purchase is bought once and kept forever, across the user's devices: a Pro unlock, an ad remover, a level pack.
• StoreKit 2StoreKit 2 is Apple's Swift API for in-app purchases and subscriptions: load products, buy, verify transactions on device, and check entitlements with async/await..
• SwiftUISwiftUI is Apple's declarative UI framework: you describe views as functions of state and it updates the screen when the state changes.
• TestFlightTestFlight is Apple's beta channel: internal testers (your team) get builds right away, external testers (up to 10,000) after a short beta review..
• WidgetKitWidgetKit is Apple's framework for Home Screen and Lock Screen widgets: small SwiftUI views refreshed on a timeline the system controls..
• XcodeGenXcodeGen generates an Xcode project file from a short project.yml, so the project is readable text instead of a hand-edited .xcodeproj..

Steam and games

• DLC (downloadable content)DLC is extra content sold for a game you already have: expansions, cosmetics, soundtracks or supporter packs.
• Steam capsule imageCapsules are the store art Steam shows in lists, search, the front page and the library, each at a fixed pixel size (for example the 462x174 small capsule and the 920x430
• Steam Next FestSteam Next Fest is a week-long event, held several times a year, where unreleased games offer free demos and run livestreams, featured across the Steam store..
• Steam visibility rounds → full pageSteam's rounds of free store exposure for a new game.
• Steam wishlistA Steam wishlist is a player saving a game to buy later.
• SteamDBSteamDB is an independent database that tracks every Steam app's updates, prices, player counts and store changes from public Steam data.
• SteamworksSteamworks is Valve's developer side of Steam: the partner site for store pages, builds, pricing and events, and the SDK for achievements, cloud saves and multiplayer..

Developer basics

• CI (continuous integration)CI is a server that builds and tests your code on every push.
• Cron expressionA cron expression is five fields, minute hour day-of-month month day-of-week, that say when a scheduled job runs.
• Exponential backoff → full pageWaiting longer after each failed retry, plus jitter so clients do not retry in sync.
• HTTP 429 Too Many Requests (rate limit)HTTP 429 means you sent too many requests in a time window.
• Jitter → full pageRandomness added to retry delays so many clients do not retry at the same moment.
• JWT (JSON Web Token)A JWT is a signed token in three base64url parts, header.payload.signature, defined in RFC 7519.
• Snowflake IDA Snowflake ID is a 64-bit ID that sorts by time: a 41-bit millisecond timestamp, a 10-bit machine ID and a 12-bit sequence.
• TOMLTOML (Tom's Obvious Minimal Language) is a config file format of keys, values and [tables].
• Unix timestampA Unix timestamp counts seconds since 00:00:00 UTC on January 1, 1970, ignoring leap seconds.
• UUIDv4 → full pageThe random UUID; why databases prefer v7 for primary keys.
• UUIDv7 → full pageA time-ordered UUID; how it differs from random UUIDv4.
• x402 → full pageAn open standard that revives HTTP 402 Payment Required so APIs and agents can pay per request.