Skip to content

SAKO 0.7.1

Latest

Choose a tag to compare

@Thakay Thakay released this 20 Oct 00:33
Immutable release. Only release title and notes can be modified.

The method, the reference pages, the recipes, the glossary, the skill and the README describe what 0.7.0 added, and a test that failed on native Windows passes there. Runtime behavior is the same as in 0.7.0. SAKO remains alpha software.

From your project's Git repository, install with uvx sako@0.7.1 init, or update an existing installation with uvx sako@0.7.1 init --update.

To update from 0.7.0, run init --update between working sessions, in the main checkout, then run verify when a check is configured, since the new version makes every recorded pass stale. From 0.6.0, follow the steps in the 0.7.0 release first.

Changed

  • SAKO.md describes edit, the session's own rows and its peers' progress, named checks, a pass from another worktree and the advice for a file rewritten after its close. It also says a receipt cites a lasting home, an unmet part of a done-when becomes a new task, task IDs come from add's output, and an edit after a close is new work.
  • The limits, validation and compatibility pages cover the same additions and their limits, with fresh counts.
  • The intake recipes add a review's findings, the owner's own acts, a row moved from another ledger and an act only one session may do at a time, and show how to capture a new task's ID. The glossary defines Marker, and the skill names edit.
  • README's principles section is "Simple, not small", and README names edit, named checks, peers' progress and a pass from another worktree where a reader needs them. The tracker recipe parks rows with edit.

Fixed

  • A test of the start context compared the header's middle dot, which native Windows reads back in its code page, so the test failed there.

Validation

  • Both suites pass on Python 3.10 and 3.12 locally, with 277 tests and 22 scenarios; these counts overlap.
  • Natively on Windows from Git Bash, with Python 3.14.4 and Git for Windows 2.52.0, 277 tests ran with 46 skipped and the scenario suite passed 17 and skipped 5, with none failed.