Balance Team

Private

Posts

  • RE: Problems with 2026.11.0?

    The sound error FAQ is outdated. We now know the cause of the bug being that the engine causes memory corruption when using an unsupported sound configuration. What the FAQ's troubleshooting covers is:

    • Disabling surround sound and enabling stereo, which is a supported configuration
    • Adjusting audio hardware/software configuration to shuffle memory hoping the memory corruption doesn't land somewhere where the game crashes once it reaches there
    • Turning off sound entirely (/nosound) so the bugged code is never reached

    Your crash address 0x00915402 pops up for other sound error victims, so it is likely the cause. We fixed a different common crash/desync bug so this is the only remaining one afaik.

    Another way to fix it right now is to play on the new engine, which fixes the sound bug. You can do this by downloading the alpha client and selecting FAF Develop when creating a new game.
    For others in the future, I expect the next stable client update will support launching the new engine, rust client should get an update for it soon, and eventually we'll get the new engine on normal FAF.

  • RE: Problems with 2026.11.0?

    On Discord I haven't seen an uptick of reports that would indicate a common error with the update.
    Is it your game or client crashing? If it's the game then make sure the FAF Debugger in the client settings is enabled and post the game logs (both can be found through the client's top left menu button). The debugger will write a stack trace at the end of the game log when enabled and the game crashes.

  • RE: SC: FA (Moho) engine sources

    FAF develop has some connectivity improvements by adjusting network parameters to make use of more capable modern networking: https://github.com/FAForever/fa/pull/7364 That's the most immediate change I know we can test right now. I don't know what's the state of the other WebRTC-based netcode or ice adapter changes.

  • Cybran land buff proposals
    1. Buff brick DPS 295 -> ~304, returning 50% of the DPS it lost after the last nerf (down from 312.5) because:

      • the nerf's goal was to make brick weaker vs Percy, and now Percy is strong vs brick, so there is space to buff the DPS
      • the nerf's goal was to make brick weaker vs othuum/harb, and while in a 1-unit-composition matchup this nerf made it more fair per mass, the brick needs to be more unfair because their lack of shields makes sniper DPS and Ythotha AoE very strong. Their stats also struggle to beat GC/ythotha + harb/othuum compositions a bit too much, which makes getting Megalith too difficult. Monkeylord is not good against GC/Ythotha at all, so mega is required to fight those T4s efficiently without just winning hard in T2 and getting an overpowering army at later T3.
    2. Add the Pulsar, which is the unreleased cybran T3 long range anti shield unit. It was wanted by old balance team, then it got delayed and eventually canceled by Rhaelya looking for an alternative solution before adding new units. Since no alternatives have been proposed and I don't see what would work better, I think the unit should finally be added. It solves the shieldless arty's low DPS against snipers and the struggle against mid-late gun coms that OC all your bricks if you try to attack anything while easily tanking with shield upgrades and mobile shields.

    3. Make firebeetles jump at enemies.

      • it looks cool and fits the theme
      • makes them hit in the middle of unit clumps more consistently, including on attack move/attack orders.
      • reduces pathfinding needed
      • increases their effective attack range

      The jump range in my prototype is around 5, which is the current detonation trigger range, and is small compared to any gun (LABs have 14 range).
      I think after the last nerf beetle is pretty good for gameplay but is a little weak because it dies very easily, especially at T3 stage, if you don't have basically equal unit counts and are in range to tank for the beetles, so that's why I'm interested in trying this new mechanic as a straight buff.

  • RE: ACU Walk Animations – Adjustments as a Test Mod

    What I meant is that since it modifies the bone positions even slightly, it probably eventually causes desyncs if two players have different animations. It's not a big gameplay difference in your animation example, but it's a potential desync so it can't be allowed. To relate back to cheese's post, you can't have "users could have whichever animation or mix animations they like," it would be limited to what FAF implements and officially distributes.

  • RE: ACU Walk Animations – Adjustments as a Test Mod

    Animations are not safe because they modify bone positions and thus things like transport attachment position or projectile creation position.

  • RE: Hotkeys

    I merged a fix for it for next patch.

  • RE: Updated fix to audio bugs?

    I'm not sure what you mean by super buggy audio, but it would be helpful if you could try these specific workarounds since I've developed fixes for them:

    • Change from 5.1 (side speakers) or 7.1 audio to stereo. Side speakers because FA supports the older 5.1 back speakers channel format. This is the bug fixed in the binary patch.
    • If your PC uptime is more than 24.86 days, sounds will play once per game session. Restarting your PC will fix this. This is a bug in XACT so it doesn't fit in a binary patch.
  • RE: Plans for the near future?

    FAF is all volunteers so there isn't really a definitive roadmap. Current trends are possible UI improvements (being careful of automation/making things too easy) and a lot of engine bug fixes and performance improvements.
    New units and game mechanics are for mods to do due to FAF's preservation mission, but imo FAF does have some places to improve mod support in. That does have the challenge of preserving support for old mods though.

  • RE: Desync Postmortem

    @JaggedAppliance said:

    Yes these desyncs and crashes affected replays as well as live games. Actually they were really common in replays which helped a lot with testing.

    The reason why this fix was so difficult is because it's a memory corruption writing into random space. With the way your computer set up the game's memory it caused regular desyncs and crashes in replays. For a lot of other people it rarely did anything. So for Brutus's idea, this specific bug would be caught by luck in a simulation CI.

    Here's how I see potential improvements compared to my experience:

    1. Desync tracking would automatically and seriously tell devs of deysncs. I only noticed desyncs from constantly checking forums and chats, and eventually I had to go through a specific replay search to even have any data about desyncs, and even then there wasn't a really big response. Automation would've been a more serious alert and it could've even caught it much earlier on.
    2. The reason this specific patch wasn't caught early was because there was 0 testing in prod. It's a tricky bug, so fair enough that some testing in dev wouldn't catch it, but since we released it without any feature using it, it took all the way up until the painting feature for its effects to show, which made the change that introduced the bug disconnected from the updates exhibiting the bug.
    3. With AI we can do much better reviews. This bug was a very human error that is easily picked up by AI or I guess some linter. I don't think we do very complex patches most of the time, so static analysis should work quite easily. In this sense I think we could actually speed up engine patches because of their more reliable code, but of course it's best if we just have the backup testing in prod with the two points above.