Yat Ho
Yat Ho
> There seems to be a problem with the endgame now, at least among the few torrents I tried that had multiple peers. Get's stuck at ~99.5%. Huh, that's new....
@reardonia ~#7027 is ready now, the problem with endgame is fixed.~ Edit: Found more bugs. For 7026 however, I need to rethink it, it's a no-go IMO in the current...
I consider this a failed attempt, so I don't plan on touching it any further. I am only keeping it here for the record before someone come up with something...
@reardonia With some recent developments, I think this PR is viable again. Would you be interested to try out https://github.com/tearfur/transmission/commit/00eaab96cdcf993aea14500588095b2dedfecf02? It is this PR rebased as-is on top of some...
@reardonia I don't think so, right now this is all that's not merged into `main` in my `wishlist-perf` branch.
I can see that it's possible the session thread update the `latest` timestamp after the macOS app thread recorded the `now_sec` value in `tr_torrent::stat()`, causing the assert to fail. Is...
Well, yeah I guess we can be sure whatever's left in the buffer at that point can be thrown away, so I won't oppose it anymore. Then maybe add some...
> how do we test the can_read() loop (in peer_msgs and handshake) through multiple packet exchanges? Actually just splitting a `sendB64ToClient()` call into 2 can do that, it's just that...
> Can I just add you as co-author for the whole set? Sure thing.
@reardonia There's merge conflicts.