wraitii

The RM scripts sometimes contain a lot of logic to avoid these issues, but it remains tricky.

I think it's likely you didn't recompile? I should have specified this was necessary. Now there has been an auto build, so you should be able to tweak things. It is definitely working for me, see the following comparison screenshots. This is 2 groups of 100 units crossing each other's path at 0 extension, 2.5 extension, 5 extension, and then at 2.5 with lower MinPushing factors. This is ordering these 200 units to go somewhere at 0/2.5/5 Edit -> BTW, you can 'hotload' your changes by quick saving and quick loading.

Good news still on this front: I've increased the flexibility of the system so you can play with 3 more settings, and fixed an issue so it's possible to have units push each other more, so things should death ball _less_. See https://code.wildfiregames.com/D4098 which I've merged. I'll still take player input during Feature Freeze, but I expect these new settings will be overall more satisfactory.

https://code.wildfiregames.com/D4097

Less but not none, because moving units can't cross through non-moving units, so chokepoint can remain relatively crowded and ultimately impassable. And their ability to be turtled and ranged by more units remain Also, conversely, artificial choke points because of bad pathfinding feel terrible. That is a good point, but the mosh pit design that we had already made this somewhat difficult. --- Overall, I agree that the pushing feature can and should ideally be tweaked and improved, but I believe the good offsets the bad. We'll see how people feel in testing.

Both Turn ½ OOS are probably explained by different revisions, as the "mainland" map had been edited on July 4. I fully reproduce Yekaterina's commands.txt using rP25682. However, I found and fixed a potential OOS on rejoin this morning, so that might still explain the 6K turn OOS.

Just to clarify -> We're not going with the most upvoted suggestion. The poll is mostly here to see if some names get a clear lead. The poll closed on June 1st because that was the original FF date, I've reopened it up until Sunday (didn't notice that). To be honest, so far, Y seems like it's lacking in very relevant names. I'm not sure what's getting picked at all at this stage.

@Palaiologos can you share the replay? We need at least the command.txt

Would be great if we could have the OOS dump of the OOS players as well for comparison

I've updated https://code.wildfiregames.com/D3679 following discussions there.

The logs should be OK to share, but the config file contains your lobby password (hashed, but it's still usable to connect with your account).

Thanks for bringing this up. I guess it's another time where it's urgent to do nothing. I guess this might make it easier to move away from Phabricator, as people might share migration scripts. Might also end up with a straightforward fork that's not a one-man project and end up with a more functional product. We'll see.

Right. I don't see the A24 situation as that problematic either (or at least I don't really mind), so the change isn't horrible to me, merely unfortunate. Note that it's still hardcoded in C++ that idle units are allowed to bunch up together more than walking units, I could make these settings different so we can tweak the parameters individually.

This is mentioned in the very first post in this thread. I don't really like it, but I also don't think it's a very realistic situation outside of toy examples. I also don't expect it to be a particular balancing problem, but maybe I'm incorrect on that account. The 'pushing distance' can easily be modified by changing pathfinder.xml if you want to test out other settings.

See also the discussion at https://code.wildfiregames.com/D3686

Committed, thanks for reporting all

Should be fixed by https://code.wildfiregames.com/D4066 if someone else can confirm

Mh, yes, this is working as designed actually. I figured it was better to show cheats are always enabled than to not show it at all like in A24. I can either hide it again, or make it configurable.

Can you upload you matchsettings.json in the user config folder? Then delete it, that will fix the problem.

Mh, given the wide range of options we have, this does seem like something I'd be open to. Think elexis was a bit too strict there.

Thanks @Nobbi, very insightful comment Right. The latter has a patch for it, not the former. You can no longer train horses at the barracks, so I think we should just change the barracks here, for consistency. The decay is on purpose, though whether it should be changed is an open question. Not sure what you mean about train time. Yes Made a ticket for that, it's less trivial than other things. You are indeed incorrect, there are no other bonuses in A24/25 (so far) Mmh, that sounds like a bug indeed. I don't necessarily disagree, but I think removing towers was a good move and we don't really have more than 3 town-phase buildings as things stand, so it's simply not possible.

I largely disagree. While translation freeze means it's completely fair game, the chances that we do major change after feature freeze are low, and as it turned out for A24, translators usually have quite a bit of work to do anyways. And if tweaks are made, they'll usually be small tweaks that Transifex helps with fixing. Packagers, aye, players, maybe, but 'testers', certainly not. After all, people play SVN. Well, the first few bundles won't be RC but testing bundles, as per the above, so it's logical to not release it 'into the wild' too much. Commits are actually not permitted later, unless for exceptional circumstances, such as fixing a release blocker issue.

Check out a tutorial online, like this one.

Mh, that seems rather weird. I would suggest trying to get a more complete callstack by running in a debugger.

If there is a specific topic for A25, please start a thread in the balancing discussions subform or make a diff.