muesli / reflow

100%

master: 100%

LAST BUILD ON BRANCH refactor/go-runewidth-to-uniseg

branch: SELECT

CHANGE BRANCH

x

• No branch selected
• compressor
• dependabot/github_actions/actions/checkout-3
• dependabot/github_actions/actions/checkout-4
• dependabot/github_actions/actions/setup-go-3
• dependabot/github_actions/actions/setup-go-4
• dependabot/github_actions/golangci/golangci-lint-action-3
• dependabot/go_modules/github.com/mattn/go-runewidth-0.0.14
• dependabot/go_modules/github.com/mattn/go-runewidth-0.0.15
• github-workflows
• master
• refs/tags/v0.3.0
• v0.1.0

Build # 7703176316

Build Type

Pull #71

github

Committed by mikelorant

Commit Message

Replace go-runewidth with uniseg

Replace the use of `RuneWidth` and `StringWidth` from `mattn/go-
runewidth` with equivalent functions from `rivo/uniseg`.

It is important to be aware that using `RuneWidth` will not be accurate
as the width of a rune cannot be determined in isolation. This requires
a shift to thinking about grapheme clusters instead.

Unfortunately due to the complexity of identifying grapheme clusters,
there has been some signifcant performance regressions in two
functions:

- PrintableRuneWidth: 10x slower
- TruncateString: 4x slower

Two other functions have had performance improvements:

- MarginString: 2x faster
- PaddingString: 2x faster

The documentation for `rivo/uniseg` mentions the use of `Step` and
`StepString` performing "orders of magnitude faster" than using the
`NewGraphemes` method. However, implementing these changes only
resulted in a 10% performance increase.

Signed-off-by: Michael Lorant <michael.lorant@nine.com.au>

Pull Request Pull Request #71: Replace go-runewidth with uniseg

Run Details

576 of 576 relevant lines covered (100.0%)

28.59 hits per line