All pull requests

Pull requests

• fix(tar): reject compressed tar archives

  #2186·dmpetrov opened Nov 6, 2026Contributor·

  ·
  14

• feat: read LeRobot v3 datasets

  #2184·dmpetrov opened Nov 6, 2026Contributor·

  ·
  12

• feat: read MCAP robotics logs

  #2183·dmpetrov opened Nov 6, 2026Contributor·

  ·
  18

• feat: add read_webdataset

  #2182·dmpetrov opened Nov 6, 2026Contributor·

  ·
  15

• perf(video): read with 1 MiB blocks when decoding from mid-file

  #2181·dmpetrov opened Nov 6, 2026Contributor·

  ·
  13

• fix(file): give a buffered stream from File.open() without the cache

  #2180·dmpetrov opened Nov 6, 2026Contributor·

  ·
  21

• fix(hf): read pinned files at their revision

  #2177·dmpetrov opened Nov 6, 2026Contributor·

  ·
  7

• fix: to_storage fails on a signal that holds no files

  #2176·dmpetrov opened Nov 5, 2026Contributor·

  ·
  8

• fix: to_storage copies image and audio files byte for byte

  #2175·dmpetrov opened Nov 5, 2026Contributor·

  ·
  8

• perf: download remote files to local paths with one GET

  #2174·dmpetrov opened Nov 5, 2026Contributor·

  ·
  14

• refactor: make the listing name decode pattern public

  #2172·amritghimire opened Nov 5, 2026Contributor·

  ·
  3

• fix(cli): show usage when dataset subcommand is missing

  #2169·bosprimigenious opened Nov 4, 2026Contributor·

  ·
  1

• fix: include root files in literal globstar matches

  #2168·bosprimigenious opened Nov 4, 2026Contributor·

  ·
  1

• fix(delta): require unsafe opt-in for slicing

  #2160·bosprimigenious opened Nov 3, 2026Contributor·

  ·
  1

• fix(cli): report size for single-file du

  #2153·bosprimigenious opened Nov 2, 2026Contributor·

  ·
  1

• perf: budget native worker threads and shrink trailing UDF tasks

  #2147·dmpetrov opened Oct 31, 2026Contributor·

  ·
  34

• perf: materialize computed SQLite join keys and simplify path functions

  #2146·dmpetrov opened Oct 31, 2026Contributor·

  ·
  43

• fix: bound prefetch queues while preserving batch units

  #2145·dmpetrov opened Oct 31, 2026Contributor·

  ·
  29

• perf: prepare scalar UDF inputs once and skip unused prefetch

  #2144·dmpetrov opened Oct 31, 2026Contributor·

  ·
  33

• feat: install a task-oriented DataChain skill by default

  #2143·dmpetrov opened Oct 31, 2026Contributor·

  ·
  62

• perf: avoid per-element conversion of scalar vectors

  #2142·dmpetrov opened Oct 31, 2026Contributor·

  ·
  32

• Don't save listings of local directories in Studio

  #2136·dreadatour opened Oct 31, 2026Contributor·

  ·
  13

  

• feat: add dc.describe() and dataset describe CLI

  #2134·dmpetrov opened Oct 31, 2026Contributor·

  ·
  41

• perf(tabular): store plain rows as tuples, with no model per row

  #2132·shcheklein opened Oct 30, 2026Contributor·

  ·
  29

  

• perf: read media metadata without filling the file cache

  #2118·dmpetrov opened Oct 28, 2026Contributor·

  ·
  76