Eyevinn / go-608 / 30949712146
85%
main: 85%

LAST BUILD BRANCH: chore/prepare-v0.10.0-dev
DEFAULT BRANCH: main
Ran
Jobs 1
Files 37
Run time 1min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
coverage: 85.398%. Remained the same
30949712146

Pull #74

github

tobbee 
chore(release): v0.9.0

- internal/version.go: v0.8.0 -> v0.9.0, and commitDate set to this commit's timestamp
- CHANGELOG.md: date the [0.9.0] section, open a fresh [Unreleased], and add the v0.9.0 compare/tag links

Minor rather than patch: generate gains the two direct-write caption modes (paint-on and roll-up, per frame and per unit), and two changes alter emitted output for every caller -- the default UTC line drops its date, and NumCues divides down instead of rounding.

commitDate stops being empty. It has been "" since the repo was scaffolded, so an un-stamped go run reported a bare v0.9.0 while only Makefile builds carried a date; both defaults now come from the release commit, and go608-clock -version reports 'v0.9.0, date: 2026-08-04' either way. Stamping still wins where it is applied, and since GetVersion renders only the date, later commits on the same release do not move it.

Note that the v0.9.0-dev opener was skipped after v0.8.0, so version.go had been reporting v0.8.0 for the nine commits since. The next cycle should open with chore: prepare v0.10.0-dev to restore the pair.

Corrections found while checking the release notes against v0.8.0, all of them claims I had written earlier in this cycle and had not measured:

  - The Fixed section claimed the continuous pop-on generator overran at 23.976 fps
    before the width change. It did not: the 23-pair build plus its EOC is exactly the 24
    pairs that rate allows. Section removed -- nothing in 0.9.0 fixes a defect that a
    released version had.
  - 'Five mode/policy combinations did not fit' was an undercount from testing only two
    of the three roll-up window sizes. RU2/3/4 all cost the same single mode-entry pair,
    so all three failed at both 25 and 23.976 fps. Restated without a count.
  - 'Roll-up costs two extra pairs per line over paint-on' was wrong in both README and
    WithRollUp. Measured over 1-4 lines, roll-up costs L-1 more: its overhead is 1+L
    (mode entry ... (continued)
Pull Request #74: chore(release): v0.9.0

3819 of 4472 relevant lines covered (85.4%)

0.99 hits per line

Jobs
ID Job ID Ran Files Coverage
1 30949712146.1 37
85.4
 GitHub Action Run
Source Files on build 30949712146