proposal: slices: Add Find

[icholy]

Proposal Details

A function for finding the first value in a slice which satisfies a predicate function.

package slices

// Find returns the first value satisfying f.
// The second return value reports whether a value was found.
func Find[S ~[]E, E any](s S, f func(E) bool) (E, bool) {
	for _, v := range s {
		if f(v) {
			return v, true
		}
	}
	var z E
	return z, false
}

[ianlancetaylor]

We already have https://pkg.go.dev/slices#IndexFunc.

[icholy]

I have this code:

var prev *github.CheckRun
for _, r := range res.CheckRuns {
	if r.GetName() == "mergelock" {
		prev = r
		break
	}
}

Using slices.IndexFunc looks like this (which is the same number of lines and harder to read):

var prev *github.CheckRun
prevIdx := slices.IndexFunc(res.CheckRuns, func(r *github.CheckRun) bool {
	return r.GetName() == "mergelock"
})
if prevIdx >= 0 {
	prev = res.CheckRuns[prevIdx]
}

It would be nice if I could write:

prev, _ := slices.Find(res.CheckRuns, func(r *github.CheckRun) bool {
	return r.GetName() == "mergelock"
})

[gabyhelp]

Related Issues

• proposal: x/exp/slices: Find function to select first element matching a condition #52006 (closed)

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

[ianlancetaylor]

To make any argument for adding this function when we already have IndexFunc, it would be good to see examples in existing Go code (code that you didn't write) where it could be used to simplify the code. Thanks.

[icholy]

Here's an example:

go/src/crypto/tls/key_agreement.go

Lines 170 to 180 in 37f27fb

	var curveID CurveID
	for _, c := range clientHello.supportedCurves {
	if config.supportsCurve(ka.version, c) {
	curveID = c
	break
	}
	}
	if curveID == 0 {
	return nil, errors.New("tls: no supported elliptic curves offered")
	}

This could be rewritten as follows:

curveID, ok := slices.Find(clientHello.supportedCurves, func(c CurveID) bool {
	return config.supportsCurve(ka.version, c)
})
if !ok {
	return nil, errors.New("tls: no supported elliptic curves offered")
}

Here are some others:

go/src/debug/pe/file_test.go

Lines 504 to 513 in 37f27fb

	var bss *Section
	for _, sect := range f.Sections {
	if sect.Name == ".bss" {
	bss = sect
	break
	}
	}
	if bss == nil {
	t.Fatal("could not find .bss section")
	}

go/src/runtime/pprof/protomem_test.go

Lines 216 to 225 in 37f27fb

	var s *profile.Sample
	for _, sample := range p.Sample {
	if sampleToString(sample) == expectedSample {
	s = sample
	break
	}
	}
	if s == nil {
	t.Fatalf("expected \n%s\ngot\n%s", expectedSample, strings.Join(profileToStrings(p), "\n"))
	}

go/src/os/stat_test.go

Lines 99 to 109 in 37f27fb

	var lsfi2 fs.FileInfo
	for _, fi2 := range fis {
	if fi2.Name() == base {
	lsfi2 = fi2
	break
	}
	}
	if lsfi2 == nil {
	t.Errorf("failed to find %q in its parent", path)
	return
	}

go/src/cmd/link/internal/ld/elf_test.go

Lines 357 to 370 in 37f27fb

	var wsro *elf.Section
	for _, s := range elfFile.Sections {
	if s.Name == wsroname {
	wsro = s
	break
	}
	}
	if wsro == nil {
	if k == 0 {
	t.Fatalf("test %s: can't locate %q section",
	test.name, wsroname)
	}
	continue
	}

go/src/cmd/link/internal/ld/dwarf_test.go

Lines 492 to 501 in 37f27fb

	var iEntry *dwarf.Entry
	for _, child := range childDies {
	if child.Tag == dwarf.TagVariable && child.Val(dwarf.AttrName).(string) == "i" {
	iEntry = child
	break
	}
	}
	if iEntry == nil {
	t.Fatalf("didn't find DW_TAG_variable for i in main.main")
	}

go/src/cmd/internal/objfile/goobj.go

Lines 48 to 54 in 37f27fb

	var arch *sys.Arch
	for _, a := range sys.Archs {
	if a.Name == e.Obj.Arch {
	arch = a
	break
	}
	}

go/src/cmd/go/internal/work/buildid.go

Lines 279 to 288 in 37f27fb

	compiler := ""
	for _, line := range lines {
	if strings.HasPrefix(line, " ") && !strings.HasPrefix(line, " (in-process)") {
	compiler = line
	break
	}
	}
	if compiler == "" {
	return "", "", fmt.Errorf("%s: can not find compilation command in %q", name, out)
	}

[Agent-Hellboy]

I don't have much experience, but I don't think we should bloat the standard libraries or add new built-in function, just for cleaner code. Look at the heap library, I'm becoming fan of Go because of its simplicity.

All of the example code you have shared from go code base is pleasing to my eye. There is no need of Find

[seankhliao]

see previously #52006 (comment)

[icholy]

That is, there is still a required result check, and it still is fundamentally a statement. The win is entirely not repeating x[ ] around the result.

This is not correct. In a lot of cases the check is unnecessary and may be omitted.

[ianlancetaylor]

@icholy Thanks for the example, but I have to say that if somebody sent me a patch to change the current code in key_agreement.go to use your suggested replacement, I would reject the patch. The original code is longer but to me it seems clearer.