LoginSign Up
← Back to all sparks
S

samplr

INFRA · APIS
Velocity0.0

Compare Human Performance to Sampling Algorithms

github.com/lucas-castillo/samplr ↗

A cognitive-science sampling package ships once, then goes quiet for eighteen months

cognitive-sciencesampling-algorithmsmcmcdormantr-package
◆Current state
samplr compares human performance against sampling algorithms, giving cognitive scientists the MCMC machinery to test whether people behave like samplers. Its entire public history is three releases: a 1.0.0 in August 2024, a floating-point fix eighteen seconds later, and then nothing until a February 2026 patch. The release notes are unusually thin even by CRAN standards.
◆Where it's heading
The feed shows a package that shipped and stopped. The 2024 tags were both created in one sitting and say almost nothing; the 2026 release is a row-count bug in Mean_Variance() bundled with citation metadata, a dropped dependency and http-to-https link fixes — the housekeeping profile of a package being kept alive for the paper that cites it rather than actively developed.
◆Prediction
Adding citation information to the README is usually the move of a maintainer expecting the package to be referenced rather than extended. On this cadence the next release is more likely another CRAN-hygiene patch than new algorithms; there is not enough in these notes to say otherwise.

◆Recent moves

  1. 5mo ago

    Mean_Variance() row count corrected after eighteen months

    The package's first release in a year and a half fixes Mean_Variance() returning more rows than there were IDs — a real defect in output shape — and otherwise trims the latex2exp dependency and tidies links. The citation block added alongside it says something about intent.

    View source ↗
  2. 1y ago

    Floating point comparison fix

    A single-issue patch published eighteen seconds after 1.0.0, both tags evidently backfilled in one sitting.

    View source ↗
  3. 1y ago

    samplr 1.0.0

    The 1.0.0 tag records a CRAN release and two author handles, nothing more. The note does not claim to be a first appearance and gives no changelog, so what the version contained has to be read off the package rather than the feed.

    View source ↗