G38.2 Weird behaviour

Did the update to 2.9.9 change behaviour of G38 ? First i blamed my cam and redid the job, but its definitely the probing. Finally had to do manual touch off.
How can i get back the behaviour pre this update?

How did you install 2.9.9? The ISO installer is still a work in progress. Not that that should matter for internal code 

This iso installation is more than 1 year old. The other day, there was an update for another software available, i did sudo apt update and sudo apt dist-upgrade. Some updates coming in, afterwards linuxcnc was 2.9.9
Since then, my long time working probing routines set z-offset somewhere (at least) above the part.
 

Now found the error.

5401-5409 - Currently applied tool length offset for X, Y, Z, A, B, C, U, V & W. Set by G43/G43.1/G43.2, cleared by G49. Volatile.

So i don't have access to toollength #5403 anymore, except i use G43/G49 to load the variables.
Thats f*** crapy update

Why i am upset?
Since version 2.5 this parameter are in use, now it changed overnight, instead creating a new set of parameter, if someone needs that. Like that it breaks configurations with probing routines and parts, when offsets are set wrong, without warning. And i am not going to set G43/G49 in probing routines. The only place to do that is in gcode. Don't want this risk for crash, when unwanted, toollength is cancelled. 

MaHa wrote:

Now found the error.

5401-5409 - Currently applied tool length offset for X, Y, Z, A, B, C, U, V & W. Set by G43/G43.1/G43.2, cleared by G49. Volatile.

So i don't have access to toollength #5403 anymore, except i use G43/G49 to load the variables.
Thats f*** crapy update

Why i am upset?
Since version 2.5 this parameter are in use, now it changed overnight, instead creating a new set of parameter, if someone needs that. Like that it breaks configurations with probing routines and parts, when offsets are set wrong, without warning. And i am not going to set G43/G49 in probing routines. The only place to do that is in gcode. Don't want this risk for crash, when unwanted, toollength is cancelled. 

Can anyone else confirm this?
And more importantly, if true, how this happened?

This change was made in response to a long standing bug report:
github.com/LinuxCNC/linuxcnc/issues/2994

I have opened an issue about this on GitHub:
github.com/LinuxCNC/linuxcnc/issues/4216

OK, thank you Aciera.

The old behavior did not respect the old docs, the old docs also specified that the parameter are set with G43

Tool Offsets for X, Y, Z, A, B, C, U, V & W. Set by G43. Volatile.

I guess there maybe many subroutine codes that relied on the old faulty behavior, even when the docs state you should set them with G43...

But the reasoning to use a different set of parameters is strong, we have had it wrong for so long time, wrong is the new correct...

Edit:
github.com/LinuxCNC/linuxcnc/pull/4219
follow progress here, you should get 2.9.10 with this reverted change soon.
The new behavior is in #5081-#5089

I have deleted the debs from the server (as much as I could do while at work) and will work on getting an update out this evening (GMT+1)

Thank You all, for reply and taking care of this. I wrote already, why this update leads to problems with my workflow.
Sorry it's me, sometimes complaining after a new release, but i like the system running rocksolid and safe - then i have a relaxed time, operating the machine.
About new releases, its always great work with all that fixes and improvements.