[3.13] gh-117174: Adapt test_multiple_statements_fail_early to new REPL behavior (follow-up gh-131065) (GH-131836)

[miss-islington]

Adapt test to new REPL behavior (follow-up gh-117174)
(cherry picked from commit a6cf827)

Co-authored-by: Bartosz Sławecki [email protected]

[Yhg1s]

@bswck unfortunately it doesn't look like the fix in GH-131836 was enough to fix the failures in 3.13.

[bswck]

Looking at it

[bswck]

The error isn't relevant to this fix, but to the fix in gh-131065, which was not backported.
It must have been overlooked that the test does exist in 3.13. Backporting gh-131065 should solve this problem.
Hm, no, I got lost.

[bswck]

Oh, I see.

gh-129488 added two new tests to the main branch:

• (3.14) Lib.test.test_pyrepl.test_interact.TestSimpleInteract.test_multiple_statements_fail_early
• (3.14) Lib.test.test_repl.TestAsyncioREPL.test_multiple_statements_fail_early

They started failing upon changes related to gh-117174, which is why Pablo created gh-131065, which was merged 3 weeks ago.

gh-129488 has been backported to 3.13 in gh-130513 very recently, readding these to the 3.13 branch:

• (3.13) Lib.test.test_pyrepl.test_interact.TestSimpleInteract.test_multiple_statements_fail_early
• (3.13) Lib.test.test_repl.TestAsyncioREPL.test_multiple_statements_fail_early

Because the backport readded the latter test regardless of gh-131065, the buildbot reported failures for it.

The reason this still fails is because I assumed that gh-130513 had been merged long ago, so I thought that gh-131065 wouldn't have to be manually applied to 3.13. So instead, I fixed the first test, which had never been an issue, but the actual fix (gh-131065) was never applied to 3.13, because gh-130513 hadn't been merged back then.

The simplest way to get out of this situation is to cherry-pick gh-131065 to this PR.
It's not bad that gh-131836 was merged to main (just unnecessary?), but we also need to reapply gh-131065 here.

[bswck]

@Yhg1s I don't have permissions to write to backport branches. Could you please cherry-pick/reapply gh-131065 in this PR?

[bswck]

Ah, and the most important part: conclusions. Before merging an old PR, it's good to re-run CI on it after refreshing it. Will remember to do that with my PRs 🫡

[Yhg1s]

GH-131850 is the backport of the other fix.

[Yhg1s]

@bswck Thanks for sorting out the test failures :)