After further debugging, I see this is related to the Inc value you calculate. Given some further analysis, to me as a newbie of that format, it looks like the Inc value is especially important in cases with observers.
That being said, I added another check for only a case with 2 players and without observers. In that case, the replay parses correct.
I also tested it locally with a couple of other replays and various settings. I couldn't find a regression, however, I`d really like your pair of 👁️ on it.
Hey @scopatz 👋
We have just hit another replay where the winner detection got wrong. You can find that replay here:
https://creepcamp.de/media/replays/RBTV%20Liga%20Season%207/Liga%204%20E/IHartmanI_vs_ENA1337_Echo_Isles.w3g
After further debugging, I see this is related to the Inc value you calculate. Given some further analysis, to me as a newbie of that format, it looks like the Inc value is especially important in cases with observers.
That being said, I added another check for only a case with 2 players and without observers. In that case, the replay parses correct.
I also tested it locally with a couple of other replays and various settings. I couldn't find a regression, however, I`d really like your pair of 👁️ on it.
LMK if there is something more I can do.