Proposed solution for issue #8919

[eqperes]

Returns s.len + 1 when substring sub is empty, as proposed in the issue discussion.

[timotheecour]

why not throw an exception (eg: ValueError) instead when sub is empty? seems more reasonable to me rather than the seemingly arbitrary s.len+1 (yes I read the rationale regarding replace(s, "") which also seems like it should throw).
Throwing catches bugs.

[Araq]

This should instead use assert or doAssert, it's an API misuse.

[timotheecour]

I agree with @Araq it should fail instead of silently accepting nonsensical inputs;
what about https://github.com/nim-lang/Nim/blob/devel/lib/pure/strutils.nim#L1520 case mentioned #8919 (which was rationale for the original fix that returned s.len+1)

I just tried,
replace("foo", "", "1") returns 1f1o1o1 ; shouldn't that also assert or doAssert according to same logic? would seem saner

[Araq]

I just tried,
replace("foo", "", "1") returns 1f1o1o1 ; shouldn't that also assert or doAssert according to same logic? would seem saner

This was discussed and we choose this implemementation because Python/JS/whatever also does it this way.

[eqperes]

So, it should use doAssert sub.len > 0 and only accept non-empty substrings?

[Araq]

Exactly.

[eqperes]

👍 done