2.9.9: changes to parameters 5401-5409 break existing probing routines

[Sigma1912]

As reported on the forum:
https://forum.linuxcnc.org/38-general-linuxcnc-questions/58927-g38-2-weird-behaviour#347435

While 725231f makes the parameters behave as documented it has broken the behavior of probing routines in the wild.

Maybe a new set of parameters might need to be considered instead of altering the behavior of the established ones.

[Sigma1912]

cc @grandixximo

[grandixximo]

Sure, they can be shifted over, but I don't think they should, user knows he should load the offsets if he wants them on, he just doesn't want to do it, old docs also stated to load with G43, nowhere it ever said that it should contain the offset that will be loaded when you send G43, he was using them incorrectly IMO

[andypugh]

The way I see it, if users have found that their routines did not follow the docs, and worked around that, and now we have broken their routines, that's our fault, not theirs.
I am feeling that this is a breaking change that I shouldn't have included in 2.9.
But it was also a bug fix.
So, going forward, we probably need a policy for bug fixes that are breaking changes.
(and that is probably "leave the bug alone" unfortunately)

Anyway, the question is whether I need to re-release a 2.9.9, or a 2.9.10 or whether we try to cover this in the release notes.

Is this going to break tools, ruin parts and break machines? I suspect that it is.

Note that I haven't made a release announcement yet, though I see that RodW did on Facebook.

[grandixximo]

users already updated, so you need go for 2.9.10 otherwise they will not get the update.
I'm preparing the patch, moving over the new behavior to #5081-#5089
what about master? same thing, we are keeping the old behavior on the old parameters, and new behavior in the new parameters?
Are deprecation messages from gcode something we want to try? I thought about it, but I think better not...

[grandixximo]

@wucke13 we are putting back the bug on the default parameters, will have to use #5081-#5089 for the new behavior.

[grandixximo]

The way I see it, if users have found that their routines did not follow the docs, and worked around that, and now we have broken their routines, that's our fault, not theirs.

We should not break old behavior even when buggy, that sucks but is the reality we live in...