Repository navigation
Releases: sightmap/jev-turbo
Releases · sightmap/jev-turbo
Release list
v0.2.2
Changelog
- 195ae65 Version Packages (#22)
- dc73590 Ranked ideas 1, 2, 4, 5, 7 and 8: effect judge, titled entries, second look, check distillation, same-names, journey mining (#21)
- 204dca3 README: quick start after the intro, one results section, details in bench (#20)
- c6f3211 render-demo: ship Inter and JetBrains Mono, encode the GIF without dither
- 375ea84 README: written for a first-time reader
- 993f6e3 README: the map-vs-no-map table and the opening score carry the current IKEA runs
v0.2.1
Changelog
v0.2.0
Changelog
- b97fece Merge pull request #15 from sightmap/changeset-release/main
- 53bb777 Version Packages
- e5111cc Merge pull request #14 from sightmap/land/demo-ikea
- b600d91 Merge pull request #13 from sightmap/land/tools-mode
- ea7a82c Merge pull request #12 from sightmap/land/grow-honest
- 7c563c8 Merge pull request #11 from sightmap/land/map-score
- 0498f0d demo: captions count acted steps
- af7278a demo: the map act recorded in full
- b7ca543 changeset: demo ikea
- 2a9c1b7 README: three ways in on ikea.com
- a97cfec demo: KALLAX into the bag, three acts
- ee95a1f bench: IKEA runs, three conditions, KALLAX goal
- c034646 bench: the IKEA goal is the KALLAX shelf unit, which needs the search
- 9a63744 bench: IKEA runs, three conditions
- d2d713b bench: IKEA finish check requires the white bookcase
- e5861b8 demo: render several recordings as acts
- 64d7e17 bench: IKEA suite, map and tool layer (BILLY into the bag)
- 2220701 bench: tools mode rerun on the fixed loop
- 58f18f9 explore, jev-turbo: small tool-mode fixes
- 23535ca explore: tool steps follow the same wasted rules as element steps
- cc55416 tools: sightkick calls and builds get a 60 s bound
- 806fd4c explore: a failed tool is skipped for one pick, its guidance kept
- 78bf7c1 bench/README: tools mode, why the medians moved
- a0ad56f changeset: tools mode
- 662633b bench/README: tools mode, what moved the medians
- 3e5c28b bench: tools mode on saucedemo; books grow regenerated
- 15ce84c jev-turbo --tools: run a suite over a sightkick tool layer
- fffae32 explore: tools mode; a picked tool runs through sightkick, elements remain the fallback
- d8c07a8 explore: load a sightkick IR and offer its tools as picker options
- 964d6dc grow: docs match the role rule
- 6ced239 grow: one place that defaults the Namer
- de4693d grow: an empty Decision.Name falls back to the template
- 8fa3b08 grow: Components lists components, not views
- 50d84d7 docs: what grow writes and what it does not
- 9099d11 changeset: grow honest
- 8519e71 grow: books run with the role rule; docs say what grow is
- b2a50ab bench --grow: write the grown components and model calls into the run file
- f3fb4bb grow: Namer interface; TemplateNamer is the default
- 4e1e7d8 grow: decide obvious kinds from roles; Jev only for repeated containers
- a26a2e5 bench/README: what each score row counts and when it is not measurable
- a805168 score: rune widths and an optional picker label
- 1130b0f score: accept a single run file
- a2bd261 score: print - where a row is not measurable
- 4f45c10 score: medians keep their half
- da984e4 changeset: map score
- e77be29 bench/README: the score reads on the map when the picker is fixed
- 9c3026b jev-turbo score: compare run files side by side
- 32b1747 explore: Score folds a run file into the map's fitness
v0.1.3
Changelog
- f1c4412 Merge pull request #10 from sightmap/changeset-release/main
- ede0e63 Version Packages
- 90acf9b Merge pull request #5 from sightmap/map-vs-no-map
- cc13551 bench: flights rerun with joint confidence; numbers updated
- 111c91c README: suite sizes, what a step counts, the Enter caveat
- ace1d25 explore: pin Run.Metrics, the table tail, and the stale trigger with tests
- ba0b853 jev-turbo: --no-map and history_count in the help text
- 1635799 spec: a history_count needs min ≥ 1 and a substring
- cf4dd77 explore: confidence on a grouped pick is the joint probability
- b2bbbbc changeset: map vs no map
- 8562873 README: saucedemo median is 4.5; steps per reached goal
- bebf17c README: flights runs headless; name the two loop bugs; define the columns
- 507658a README: what the map changed, measured
- 8bda9cd bench: rerun map vs no-map on the fixed loop
- 0d894a2 explore: a control inside a same-named container is still a candidate
- 2ef4096 bench: saucedemo finish checks readable without the map; rerun
- bc8f0a8 explore: a step records the page it was picked on after a second look
- abfadba explore: look twice for suggestion options before offering the whole page
- 04b4049 jev-turbo: never bring the tab to the front; --headless with --start
- 4e090de bench: map vs no-map runs for saucedemo, journeys and flights
- 7a8f709 bench: journeys suite with two long saucedemo goals
- dbc0103 bench, explore: --no-map runs the same loop over an empty map
- 6d9a9e9 spec: history_count finish check for long goals
- 6cca63e explore: one step filter for run and suite medians
- aa288b2 explore: run and suite metrics (wasted, fallback, low confidence, candidates)
- 5ac4a10 explore: record candidates, named, confidence, fallback and wasted per step
v0.1.2
v0.1.1
v0.1.0
First release. jev-turbo explore drives a running sightmap browser session toward a goal with Jev picking every step over the annotated component tree; jev-turbo bench runs the goal suites under bench/; --grow builds the corpus while exploring. Same binaries as published to npm as @sightmap/jev-turbo@0.1.0.