Any idea why for the following test case:

$string = 'můj žlutý kůň';
var_dump(explode('ů', $string, -2));

the result is marked as NOT valid UTF-8 string? The real result is an array with single character: https://3v4l.org/oRGNM/rfc#vgit.master, maybe due some interning?

Both the delimiter and string don't have the "valid UTF8" flag. Using mb_check_encoding to attempt to add the flag doesn't work because they're both indeed interned.

Both the delimiter and string don't have the "valid UTF8" flag. Using mb_check_encoding to attempt to add the flag doesn't work because they're both indeed interned.

If the delimiter is valid UTF-8 and the string is valid as well, the result "is the same but has the delimiter cut away", ie. leading to a valid UTF-8 string. What do you exactly mean?

What do you exactly mean?

Print the flags of the delimiter and the input string and you'll see they both don't have the UTF-8 valid flag. They were interned by the compiler and the compiler doesn't set that flag. Trying to add that flag using mbstring wouldn't work either because a flag can't be added once they're interned.

Thank you for the hint. I created #10853 feature request, as the check is quite cheap, all interned/compiled strings should be checked upfront to utilize opcache.

Thank you for the hint. I created #10853 feature request, as the check is quite cheap, all interned/compiled strings should be checked upfront to utilize opcache.

You're welcome. I think Alex was going to look into something like this. I'm not sure anymore but I thought I saw him say something like that in a comment on a PR some time ago...

maybe later