GH-44846: [CI][C++] Revert change for maybe uninitialized variable to fix thread sanitizer failure

[raulcd]

Rationale for this change

The thread sanitizer build was failing since the original was merged: 2e1ddf5

What changes are included in this PR?

Revert change

Are these changes tested?

Via CI

Are there any user-facing changes?

No

• GitHub Issue: [C++][CI] ubuntu-cpp-threadsanitizer job fails consistently with a data race on asof_join_node_test  #44846

[raulcd]

@github-actions crossbow submit test-ubuntu-24.04-cpp-thread-sanitizer

[github-actions]

⚠️ GitHub issue #44846 has been automatically assigned in GitHub to PR creator.

[raulcd]

I can't reproduce locally, I am just trying to understand if a previous commit is the culprit of the issue.

[github-actions]

Revision: c58ba59

Submitted crossbow builds: ursacomputing/crossbow @ actions-6b9fcced11

Task	Status
test-ubuntu-24.04-cpp-thread-sanitizer

[raulcd]

@pitrou @kou this seems to be the culprit for the thread sanitizer failures. Should I revert the change?

[pitrou]

I don't know, I don't understand how those TSan failures could be related to the change being reverted.

[pitrou]

Also, I don't know why CI was mostly skipped on this PR, did you put "WIP" in the PR title at some point?

[raulcd]

yes, I was initially trying to understand if this was the issue or not, that's why I used WIP. I've just rebased with a new commit message to trigger CI again.

[pitrou]

Ok, I think the original commit just surfaces a lingering bug in ConcurrentQueue. UnsyncFront is simply not safe, as it calls deque::front() which can be implemented in terms of deque::begin(), but a concurrent call to deque::push_back() is specificed as invalidating all iterators.

[raulcd]

Thanks @pitrou then let me close this PR as the change is good and it surfaced the bug. I will add your comment to the thread sanitizer issue

[raulcd]

ok, I just saw you are working on the fix already, no need to add comment then. Thanks!