[rr-it]

This changes the test-case to actually use a non-interned string.

This construct actually becomes an interned string after optimizations applied in 93fd714:

$str = str_repeat("my valid string", random_int(1, 1));

php-src/ext/standard/string.c

Lines 5641 to 5643 in 93fd714

	if (mult == 1) {
	RETURN_STR_COPY(input_str);
	}

---

Further random_int(1, 1) is optimized since PHP 7:

php-src/ext/standard/random.c

Lines 171 to 173 in c12917a

	if (min == max) {
	RETURN_LONG(min);
	}

Implemented with c12917a and further refactored with b14dd85 and 4d8dd8d.

[ndossche]

Wrong commit pointed, str_repeat got optimized such that passing 1 returns the input string, that's the root cause not the ext/random changes.
Your alternative doesn't work because of SCCP.
Also, many tests rely on the old str_repeat+random_int trick. The simplest fix is likely to change the randomness bounds to 2,2

[rr-it]

The simplest fix is likely to change the randomness bounds to 2,2

@ndossche That one fixes it. Do you add a new PR or shall I?

---

Your alternative doesn't work because of SCCP.

This alternative actually works to create non-interned strings:

$strInterned = "my valid";
$strNonInterned = $strInterned . " string";

Also see:
#25371

[ndossche]

This made it into 8.6, so the target branch should be PHP-8.6. You may do the fixes in this PR.

[rr-it]

Commit rebased on branch PHP-8.6 and PR changed to target branch PHP-8.6.

[LamentXU123]

Thx! LGTM otherwise you remove the unrelated change below.

[rr-it]

remove the unrelated change below.

@LamentXU123 Sorry, I do not get the point. Please tell again.

The line itself is not unrelated, as this change doubles the string now.

[LamentXU123]

Oh I see. My bad, was a bit sick today. I mixed up the diff from another PR
Then it looks fine. I will merge it, thanks a lot.