PVG Inappropiately cuts throttle of the second to last unguided stage

[GoatQueerEnjoysChocolate]

MechJeb Version

2.15.1.0

KSP Version

1.12

Description

Whenever I try to launch a caveman style rocket in mechjeb PVG, the third stage seems to kill its own throttle and pvg seems to turn itself off despite I clearly setting it up so that it doesn't kill the throttle. I've even tried disabling autostaging, PVG still ends up turning itself and the third stage Aerobee engine shuts down before it has completed using all its fuel on that stage I am using Principia alongside a RP-1 full install on a brand new KSP install that uses SOL. I've also shared the craft file to an another person and even they are having the same problem I'm having with PVG

Replication Case

Multiple videos of testing the Beehive Caveman craft in which PVG shuts off before even completing the second to last stage burn
Latest reproduction of the bug but on youtube instead, this one with autostaging disabled
Step to Reproduce

1. Have a brand new full install of RP-1 with Principia and SOL on linux using proton
2. Set all the aerobee stages as unguided when using this craft: Decibat-II Craft link
3. Set fixed coasting to 130 seconds and set mechjeb to coast after Stage 4
4. Wait until its on Stage 1 and watch it as it shuts down before completing the burn leading to mechjeb PVG shutting down alongside the engine despite very clearly setting it to not shut down the throttle when module is disabled in the mechjeb settings menu

KSP.log

KSP.log
KSP.log backup on a site that will actually store the file for more than 24 hours

[lamont-granquist]

Yeah, if you watch on the video, it flips to TERMINAL guidance and then it stages. There's a bug where PVG grabs the final state vectors of the stage you are on when you enter TERMINAL guidance to compute the target orbital energy to exit TERMINAL guidance. It has nothing at all to do with module-disabling-does-not-kill-throttle or anything else. It is just a PVG bug where it grabs the wrong conditions and thinks it is done ahead of time.

Unfortunately, that bug is fixed in dev but is mixed in with the rest of the current mess that is PSG.

You may be able to work around it by setting the rocket to have an "early shutoff stage" of 0 and then setting a higher target, something like 170x1000 km in your example. That should cause the rocket to burn to completion and shutoff without entering terminal guidance or just keep the transition to terminal guidance in the last stage of the rocket.