Cœur

I can confirm Gary's analysis. I used main branch for a few days (post #4665), and one of my new downloads still ended up with corruption. ![Capture d’écran 2023-02-04 à...

@jawbroken that would be a completely different issue, and it's normally the system responsibility to avoid corruption on file move/copy. So yes, I really want to tackle the corrupted data...

@jawbroken I've been using that branch for my own usage for past three months. I implemented a solution that I was able to figure out. If you can optimize the...

@ckerr @tearfur I've now implemented the settings part in my second commit. I've created a new `tr_verify_complete_mode` for consistency. But since it's mostly identical to `tr_verify_added_mode`, we can alternatively merge...

@andreygursky the seeding resumes right after verification is done. Verification duration depends on how old is your hardware: - On a 10 years old Intel Mac, it was maybe taking...

@tearfur done in f18a9bd8d8b5775146dffd66c58e22106a6dec52

Yes, it's still very needed. While linked issues are marked as resolved because "frequent" corruptions got addressed, it's still possible to get corrupted data, and this pull request simply adds...

> I gave this PR more thought, and I think I need to revoke my approval. The current PR state executes the action list for a completed torrent _before_ the...

> Correct me if I'm wrong, `tr_torrentVerify()` does not wait for the verification to complete before returning. OK, correct, `tr_torrentVerify()` does: 1. `tor->stop_now()` 2. `tor->session->verify_add(tor)` This verification indeed happens in...

Hum test failure. I do not know how to address it. @tearfur do you think you can help me for it? ``` [ RUN ] TorrentMagnetTest.setMetadataPiece ================================================================= ==16627==ERROR: AddressSanitizer: heap-use-after-free...