Replay completion

[dreadnaut]

I thought checking the last second of the replay would give enough information, but it doesn't seem to work as expected:

• if you press Esc immediately after a crash, you get a replay ending with just a few (<20) zeros, but...
  
• if you wait until the game takes you back to the Evaluation, the replay will end with 4 seconds of zeros (about 80 bytes)
  

Checking for the last second being empty works then only for replays saved in the first case. I could check to see if exactly one second is empty, no more no less, but someone might be able to press Esc at the right time and get the right number of zeros at the end of an incomplete replay.

Any suggestions? :|

[Duplode]

I don't think we can do better than testing for exactly one empty second, at least through site automation. At least that would catch (nearly) all accidental submissions of incomplete replays.

(By the way, have there ever been any issues with replays cutting off 0.05s early at the end? I seem to faintly recall an old thread mentioning that happening with a replay by Alain, but that might be just my mind playing tricks.)

[USC_closure]

What about automatic penalty time detection? Combining DUP's path re-generating algorithm with RPLOnline could be useful (if it's possible somehow).

[Duplode]

What about automatic penalty time detection? Combining DUP's path re-generating algorithm with RPLOnline could be useful (if it's possible somehow).

In principle, restunts tools can be used for that, as dstien has demonstrated.

[USC_closure]

What about automatic penalty time detection? Combining DUP's path re-generating algorithm with RPLOnline could be useful (if it's possible somehow).

In principle, restunts tools can be used for that, as dstien has demonstrated.

But it would be great to see in practice.

(Wisdom for engineers: "In theory, there is no difference between theory and practice. In practice, there is.")