Boris-Barboris

Members
  • Content Count

    678
  • Joined

  • Last visited

Community Reputation

520 Excellent

4 Followers

About Boris-Barboris

  • Rank
    I'll be right happy to

Profile Information

  • Location Array

Recent Profile Visitors

11,407 profile views
  1. @Snark Vector3 is a struct and new operator on structs does not allocate on GC heap unless it's boxed.
  2. It's a hack to prevent vessel part upgrade code, intoduced by squad in 1.8.0, from crashing on custom control surfaces. Previous page (27) of this forum topic contains the rationale and sources.
  3. The purpose is to fix stock module bugs (inversion on angled wings, quaternion comparison bug, less than 1 meter from CoM bug) and to synchronize all surfaces by hardcoding their actuation speed. No, these problems are not gone. https://bugs.kerbalspaceprogram.com/issues/5172 https://bugs.kerbalspaceprogram.com/issues/21949 https://bugs.kerbalspaceprogram.com/issues/9407 https://bugs.kerbalspaceprogram.com/issues/9460 and their duplicates I need features (module reordering, regex matching) that are not present in MM. The easiest way to tackle this is to simply ignore this part, like I do for breaking grounds propeller blades based on their names. What's the part name?
  4. Just some fancy graphs of the telemetry of AA pitch angular acceleration controller during rapid hard pitch down and pitch up vertical snake manoeuvre: You can see the never-ending struggle, caused by the model that is too simple to reflect the real craft and has to be constantly updated. The core assumption is simple: AeroTorque = DynPressure * (Constant + K1 * AoA + K2 * ControlSurfacePos) This is a very useful linear formula that approximates both real and KSP airplanes on small angles of attack. If we were to plot AoA, ControlSurfacePos and AeroTorque as points in 3D space, this linear formula would imply that these points should lie on a plane. Let's see the data from KSP flight: https://i.imgur.com/pQMGsTl.mp4
  5. https://github.com/Boris-Barboris/AtmosphereAutopilot/releases/tag/v1.5.16 Reasonably stable. Dll by @Morse (thank you) bundled with AA.
  6. It's working as intended. FBW maintains zero sideslip, and the only way to maintain zero zideslip when you have rolled the craft is to yaw and change heading. Rocket mode makes your yaw similar to pitch and maintains angular velocity.
  7. Can't get this one, see if you have some error spam in logs, cause no such thing happens on my machine.
  8. https://github.com/Boris-Barboris/AtmosphereAutopilot/releases/tag/v1.5.16rc1 Try this one please, I'm interested in insufficient authority bug. The borked craft editor will have to be fixed with Morse's code, I'm afraid.
  9. I see, they switched from percent-based limiter to angle-based one.
  10. Ok, thanks, we'll have to use that if the devs won't fix that. In the meantime someone's gotta find out why are these SaveUpgradePipeline needed in the first place. These ConvertControlAuthority and ConvertControlAuthorityAxisGroup may indicate that my control surface code will have to be updated as well, since something has changed significantly enough to require upgrade code. I'll spin up windows wm some day to do this, just not today.
  11. What is that v180_ModuleControlSurface class, did they rewrite them for 1.8.0? If I derive from it instead of ModuleControlSufrace, will that help?
  12. Old dlls are unusable, on my rebuild this bug breaks SPH badly. I also had the same symptoms as this gentleman: It's probably better to just wait for a patch: too many bugs.