Disable nonlinear extrusion on unretract

[vovodroid]

Resolves #26808

Don't use nonlinear extraction on untertracts to prevent blobs at the line start, use only when any axis is moving.

[abortz]

lgtm. Thanks @vovodroid!

[vovodroid]

I guess it's better to extract axis movement check from planner to define/function to avoid double code:

  if (true NUM_AXIS_GANG(
      && !block->steps.a, && !block->steps.b, && !block->steps.c,
      && !block->steps.i, && !block->steps.j, && !block->steps.k,
      && !block->steps.u, && !block->steps.v, && !block->steps.w)
  ) {                                                             // Is this a retract / recover move?
    accel = CEIL(settings.retract_acceleration * steps_per_mm);   // Convert to: acceleration steps/sec^2
  }

If for some reasons no, at least change condition in stepper to the same (I inverted logical expression).

[vovodroid]

By a way, @abortz - why A and B factors are swapped? To be RepRap/Klipper compatible? It's so confusing to be other than mathematics notation Ax^2+Bx+C. It could be endless source of mistakes and wrong setting.

Any chance to make it mathematically correct, until not many people use it? (I was told in Marlin Discord that I'm the only person using it))).

[abortz]

Well, I use it too :)

But yeah, RepRap did it first so I copied their API -- blame them! I guess it hasn't been in an official release of Marlin yet, and I'm not personally opposed to changing the coefficient order.

[thisiskeithb]

I was told in Marlin Discord that I'm the only person using it

"No one" was an overstatement by someone. It's just not a popular option.

[vovodroid]

It's just not a popular option.

So can we make it Ax^2+Bx+C as it should be?

[thisiskeithb]

So can we make it Ax^2+Bx+C as it should be?

We don't have to match RRF's implementation 1:1, but I can see how that is beneficial.

Alternatively, the M592 - Nonlinear Extrusion Control page can be updated for clarity instead.

[vovodroid]

the M592 - Nonlinear Extrusion Control page can be updated for clarity instead.

I don't think it will help. Anyone (including myself) tests underextrusion, goes to any polynomial regression solver online, takes from there A,B,C and issue M592 A B C as he expects from mathematics.

[abortz]

Have tested PR on my Ender 3 S1, works great! Thanks again @vovodroid

[vovodroid]

I'll do soon refactoring I talked about:

to extract axis movement check from planner to define/function to avoid double code:

[sjasonsmith]

I'm going to go ahead and merge this change in, since it has been reviewed and tested by @abortz who originally authored the Nonlinear extrusion code.

[vovodroid]

it has been reviewed and tested by @abortz

I hove refactoring I did is also good (at least it works for me).

What about making polynomial Ax^2+Bx+C? Go with this?

[sjasonsmith]

Did I merge this too soon? Let me know if I need to revert it.

I’ll leave opinions on the need for further PRs to @abortz or @thisiskeithb who know more about this feature than I do.

[vovodroid]

Did I merge this too soon?

No, it's ok, just wanted to clarify.