v1.43-fix-sgtm-beta1

fix!: replace the gSecondsSinceLastFrame_WorldTime use by the
gSecondsSinceLastFrame_RealTime

This is quite a conceptual change: the physics are calculated not
depending on what we think is the inner game clock, but depending on what we
think is the external real clock. Theoretically this should be a worse choice,
as we calculate physics for the game world with its own world clock. But the
preliminary tests show a better stability of the physics simulation when the
game time is strongly slowed, when the fps are high (for example 100+), and of
course with a combination of both. My hypothesis for this better physics
simulation is that the inner game clock has a larger error margin / a
larger deviation than the real world clock, and that this causes
artifacts in the physics simulation.
Players tests will indicate if this is a choice to keep.

v1.41-rc1

fix: fix sorting algorithm - crashed the debug versions (#26)

- Fixed: 2 equal values could return true (sorting algorithms require that 2
  equal values return false). This caused a assertion error in debug builds and
  thus a crash.
- Fixed: testing equality of floats with zero.
- Simplified a test:
  knowing that dl > 0 and dr > 0,
  (cl * cr < 0) ? (cl > cr) : (dl * cr < dr * cl)
  is equivalent to:
  (dl * cr < dr * cl)

feat-completeConfig

v1.40-beta1

feat: make configurable the number of physics calculus substeps

v1.39

Same as v1.39-rc1 without residual useless compiled data

v1.39-rc1

fix: fix scaling egression introduced in fb32e328b847794a81d4d278b5df…

v1.38

fix!: stop corrupting saves (#19)

* fix!: stop corrupting saves

- This was seen only on AE. Our code for saving DHDT data through SKSE
  serialization for AE was faulty.
  We don't yet understand why, so our solution is to implement our own
  saves for DHDT.
- No save files for non-DHDT users

Co-authored-by: SesamePaste233 <[email protected]>

v1.38-beta2

fix: load and save DHDT files

v1.38-beta1

fix!: stop corrupting saves

This was seen only on AE. Our code for saving DHDT data through SKSE
serialization for AE was faulty.
We don't yet understand why, so our solution is to implement our own
saves for DHDT.

1.27-rc1

Based on 5b5c51a