gurudennis
07-15-19, 03:28 PM
How found: Rigorous play-testing with 3 separate real-life stopwatches that were independently verified to be accurate
Game version: 0.18, though also anecdotally present in 0.17
Setup: Any skirmish settings with more than one player present in the game
Repro rate: Close to 100%
Steps to reproduce: Start a game and have more than one player in it. Both the host and the guests will be affected.
Effects:
The in-game timer (the one that is activated with T and R keys) diverges from the real-world wall clock, i.e. from real time, by up to 20% (up to 12 seconds per minute) in a random direction (either faster or slower). The exact divergence varies over time within the same game, such that the in-game timer can be both slower and faster in the same game at different points in time. More "lag" that typically results from a greater number of players in the lobby typically correlates with greater timer divergence.
Near as I can tell, the in-game timer remains consistent with the in-game absolute clock. However, it is not consistent with the movement of in-game objects such as merchant vessels. This results in gross mis-estimation of target ship speed using the traditional vertical post method if the in-game timer is used to time the movement of the target ship. Conversely, if a real-world stopwatch is used, the estimation of ship speed nearly always comes out accurate and consistent.
Known mitigations:
a) Use a real-world stopwatch, or
b) Have only one player (yourself, the host) in the lobby
Special thanks: Rasputing and Obistonedkenobi who suffered through my obsessive time measurements in multiple games.
TLDR version: DO NOT use the in-game timer. Use a real-life stopwatch (e.g. the one in your phone) until the bug is fixed.
Game version: 0.18, though also anecdotally present in 0.17
Setup: Any skirmish settings with more than one player present in the game
Repro rate: Close to 100%
Steps to reproduce: Start a game and have more than one player in it. Both the host and the guests will be affected.
Effects:
The in-game timer (the one that is activated with T and R keys) diverges from the real-world wall clock, i.e. from real time, by up to 20% (up to 12 seconds per minute) in a random direction (either faster or slower). The exact divergence varies over time within the same game, such that the in-game timer can be both slower and faster in the same game at different points in time. More "lag" that typically results from a greater number of players in the lobby typically correlates with greater timer divergence.
Near as I can tell, the in-game timer remains consistent with the in-game absolute clock. However, it is not consistent with the movement of in-game objects such as merchant vessels. This results in gross mis-estimation of target ship speed using the traditional vertical post method if the in-game timer is used to time the movement of the target ship. Conversely, if a real-world stopwatch is used, the estimation of ship speed nearly always comes out accurate and consistent.
Known mitigations:
a) Use a real-world stopwatch, or
b) Have only one player (yourself, the host) in the lobby
Special thanks: Rasputing and Obistonedkenobi who suffered through my obsessive time measurements in multiple games.
TLDR version: DO NOT use the in-game timer. Use a real-life stopwatch (e.g. the one in your phone) until the bug is fixed.