[simpleqt]

Fixes #538.

Root cause

FlagSet.Parse and FlagSet.ParseAll reset f.args on every call but leave f.argsLenAtDash from the previous parse. After a parse that contained -- (delimiter position 1), a second parse of the same FlagSet without -- still returns 1 from ArgsLenAtDash(), which contradicts the field's documented "-1 if there is no --" contract in the current argument list.

Fix

Set f.argsLenAtDash = -1 alongside the existing f.args = make([]string, 0, len(arguments)) reset, in both Parse and ParseAll. That is the same reset the first field already gets.

Test

TestParseRepeatedArgsLenAtDash reuses a FlagSet: first parse with -- stops at index 1, second parse without -- must report -1 and Args() must match. It covers both entry points.

Evidence:

• RED with only flag.go reverted to base:
  git checkout 046b0d8 -- flag.go; go test -run TestParseRepeatedArgsLenAtDash -count=1 -v . ->
  flag_test.go:903: expected argsLenAtDash -1 after re-parse without --, got 1; FAIL github.com/spf13/pflag
• GREEN on the fixed tree (git checkout HEAD -- flag.go): same command -> PASS, ok github.com/spf13/pflag 0.042s.
• go build ./... exit 0; gofmt -l flag.go flag_test.go lists nothing.
• I did not complete a full go test . run on this Windows host (the full-binary invocation stalls in my shell harness), so I am not claiming a whole-suite result here - the focused regression plus go build ./... are what I verified, and CI is the whole-suite validation.

Signed-off-by: simpleqt 2015900818@qq.com

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota. Try requesting a review at a lower effort level.