cmd/fix: `go fix` and `go fix -diff` report the "reverse" results

[thaJeztah]

Go version

go version go1.26.1 darwin/arm64

Output of go env in your module/workspace:

not relevant

What did you do?

This is really a "nit" 🙈 and maybe it was by design but I noticed it when updating some of my examples; it looks like go fix and go fix -diff are inconsistent and report the exact reverse of changed made.

Feel free to close this one if this was by design! ❤️ - if it was not by design, I'm happy to look at contributing a change 👍

Without -diff, go fix reports the number of fixes applied:

fix: applied 2 of 3 fixes; 1 file updated. (Re-run the command to apply more.)

With -diff, go fix reports the number of fixes skipped;

fix: 1 of 3 fixes skipped (e.g. due to conflicts)

What did you see happen?

With -diff:

go fix -mod=readonly -diff ./... > /dev/null
# example.com/f
# [example.com/f]
fix: 1 of 3 fixes skipped (e.g. due to conflicts)
# example.com/g
# [example.com/g]
fix: 1 of 3 fixes skipped (e.g. due to conflicts)
# example.com/h
# [example.com/h]
fix: 1 of 2 fixes skipped (e.g. due to conflicts)
# example.com/i
# [example.com/i]
fix: 2 of 3 fixes skipped (e.g. due to conflicts)
# example.com/j
# [example.com/j]
fix: 2 of 3 fixes skipped (e.g. due to conflicts)
# example.com/k_test
# [example.com/k_test]
fix: 1 of 2 fixes skipped (e.g. due to conflicts)

Running the same without -diff reports the exact reverse;

go fix -mod=readonly ./...
# example.com/f
# [example.com/f]
fix: applied 2 of 3 fixes; 1 file updated. (Re-run the command to apply more.)
# example.com/g
# [example.com/g]
fix: applied 2 of 3 fixes; 2 files updated. (Re-run the command to apply more.)
# example.com/h
# [example.com/h]
fix: applied 1 of 2 fixes; 1 file updated. (Re-run the command to apply more.)
# example.com/i
# [example.com/i]
fix: applied 1 of 3 fixes; 1 file updated. (Re-run the command to apply more.)
# example.com/j
# [example.com/j]
fix: applied 1 of 3 fixes; 1 file updated. (Re-run the command to apply more.)
# example.com/k_test
# [example.com/k_test]
fix: applied 1 of 2 fixes; 1 file updated. (Re-run the command to apply more.)

What did you expect to see?

Both to report the same 😅 - perhaps a combination of the two ("applied X fixes" is a more positive message and "re-run the command" makes it (potentially) actionable); something like...

fix: 1 file updated: applied 2 of 3 fixes, skipped 1 (e.g., due to conflicts); re-run the command to apply more.

[gabyhelp]

Related Issues

• cmd/fix, x/tools/go/analysis: fix can print "applied 8 of 10 fixes; 2 files updated" and fail without actually applying any fixes #77482
• cmd/fix: -diff produces interleaved -/+ lines instead of grouped hunks #77957 (closed)
• cmd/fix: change -diff to exit 1 if diffs exist #77583
• cmd/fix: go fix failing with non zero exit code #77615 (closed)
• cmd/fix: 'go fix -inline .' says 'applied 2 of 3 fixes' but actually applies none #77750 (closed)
• "gofmt -d" is broken. (and "gofix -diff" too) #1916 (closed)

_{(Emoji vote if this was helpful or unhelpful; more detailed feedback welcome in this discussion.)}

[thediveo]

I would really want to see the reasons why fixes were skipped if any, instead of a number and a comment straight off an LLM.

[thaJeztah]

@thediveo you mean why my example failed? Or did you mean "I would like the messages printed by go fix to include the reason"? 😅

If it's because you were curious what example produced it; this was related to #78164 and #78165 (78164 has an example module uploaded as attachment if you want to try it) 😅

[thediveo]

Actually both, but I don't want to sideload the issue.