Skip to main content

Config.XOfTheDay

Path: build/Config/XOfTheDay.hs | Language: Haskell | Lines: ~499

Configuration constants for the "X of the Day" feature: paths, thresholds, and domain exclusions

Overview​

This module centralizes all configuration for gwern.net's "X of the Day" features—Quote of the Day, Site of the Day, and Annotation of the Day. Each feature displays a rotating selection of curated content on the homepage, drawn from metadata databases.

The module defines three categories of configuration: (1) file paths to the source databases and output HTML files, (2) quality thresholds that filter which entries qualify for rotation, and (3) a comprehensive domain blacklist preventing certain sites from appearing in the Site of the Day feature.

The blacklist is notably large (~460 domains) and excludes common platforms (GitHub, Wikipedia, Twitter/X), major publications (NYT, Guardian, Nature), academic repositories (arXiv, JSTOR), and gwern.net itself—the rationale being these are either too common to be interesting discoveries or would create self-referential recommendations.

Public API​

quoteDBPath :: FilePath​

Path to the quote database: metadata/quotes.hs

quotePath :: FilePath​

Output path for today's quote: metadata/today-quote.html

siteDBPath :: FilePath​

Path to the site database: metadata/sites.hs

sitePath :: FilePath​

Output path for today's site: metadata/today-site.html

annotDayDB :: String​

Path to the annotation database: metadata/annotations.hs

annotPath :: String​

Output path for today's annotation: metadata/today-annotation.html

minAnnotationAbstractLength :: Int​

Minimum character length for annotation abstracts to qualify: 2000

Comment notes this threshold yielded 3,313 qualifying entries (vs 10,046 at >500 chars).

siteLinkMin :: Int​

Minimum number of links required for a site to qualify: 3

siteBlackList :: [T.Text]​

List of domain names excluded from Site of the Day recommendations.

Internal Architecture​

The module contains no logic—it's purely declarative constants. The blacklist is a simple list literal of T.Text domain strings, designed for O(n) membership checks (or conversion to a Set by consumers).

Database File Format​

The .hs extension on database files suggests they use Haskell's Read/Show serialization rather than JSON or YAML. This allows type-safe deserialization directly into Haskell data structures.

Output Files​

The today-*.html files are pre-rendered HTML fragments included in page templates, regenerated during each build cycle.

Key Patterns​

Threshold Tuning Comments: The module includes inline documentation about threshold selection rationale:

-- at >500, yielded 10,046 on 2023-03-08; >2,000 yielded a more reasonable 3,313

This captures the empirical reasoning for configuration choices, useful for future adjustments.

Domain Blacklist Organization: Domains are listed alphabetically within chunks of ~5 per line, making diffs readable and merges less conflicting.

Configuration​

All values are compile-time constants. To modify:

SettingCurrent ValueEffect
minAnnotationAbstractLength2000Raises bar: fewer but meatier annotations
siteLinkMin3Ensures sites have meaningful presence in annotations
siteBlackList~460 domainsPrevents generic/common sites from appearing

Integration Points​

Consumers: The build system modules that generate X of the Day content import these constants:

  • Selection logic uses thresholds to filter candidates
  • HTML generators write to the configured output paths
  • Recommendation engine checks blacklist before selecting sites

Related metadata files:

  • metadata/quotes.hs - Quote database
  • metadata/sites.hs - Site recommendation database
  • metadata/annotations.hs - Full annotation database

See Also​

  • XOfTheDay.hs - Main module consuming this configuration
  • Hakyll.hs - Site generator that invokes XOTD generation
  • Annotation.hs - Populates the annotation database this module references
  • Config.Misc - Related configuration module with global constants
  • LinkMetadata.hs - Annotation database used for annotation-of-the-day
  • Blog.hs - Related daily content system for blog posts
  • sync.sh - Build orchestrator coordinating content generation