proposal: slices: add LastIndex and LastIndexFunc

[dsnet]

I propose the addition of:

// LastIndex returns the index of the last occurrence of v in s, or -1 if not present.
func LastIndex[S ~[]E, E comparable](s S, v E) int

// LastIndexFunc returns the last index i satisfying f(s[i]), or -1 if none do.
func LastIndexFunc[S ~[]E, E any](s S, f func(E) bool) int

which parallels the existence of:

• strings.IndexXXX
• strings.LastIndexXXX
• bytes.IndexXXX
• bytes.LastIndexXXX

In my use-case, I'm processing Unicode runes as a []rune and wanted to find the last occurrence of a given rune.

[earthboundkid]

This makes a lot of sense, but devil's advocate, should we just tell people to do iterutils.Index(slices.Backwards(s), v) in a world where iterutils exists? I think the tricky of it would be that even with slices.Backwards, you run into issues with matching just the value vs. the index. Probably it's worth it anyway.

[jimmyfrasche]

index doesn't make sense with iterators like it does with slices.

[ianlancetaylor]

I can see a reasonable argument that the slices package should have a version of all relevant functions in the bytes and strings packages.

This makes a lot of sense, but devil's advocate, should we just tell people to do iterutils.Index(slices.Backwards(s), v) in a world where iterutils exists? I think the tricky of it would be that even with slices.Backwards, you run into issues with matching just the value vs. the index. Probably it's worth it anyway.

slices.Backwards doesn't exist:

https://cs.opensource.google/search?q=backwards&ss=go/go:src/slices/

and according to the quoted comment, neither does iterutils. so I am not sure how much value would be in a suggestion thats not even actionable. also the implementation is dead simple:

https://github.com/golang/go/blob/go1.21.1/src/bytes/bytes.go#L146

[timothy-king]

I just ran into this. Comes up as a nice utility function when cloning the tail of a stack datastructure. I ended up rewriting it by hand.

[josharian]

I just ran into this as well, porting some code from string to []rune, which reinforces Ian's point above.