simulation

LinuxCNC does not have this facility.

I can think of partial solutions (with a stepper motor machine, simply turn off the stepper drivers, with a servo machine you would need to short-circuit the feedback loop in HAL, which is actually fairly simple to do)

But, the real problem comes when you _exit_ simulation mode. Should the actual machine rapid to the simulated finish point (that would be the default behaviour with the HAL-only solutions I mentioned). If the machine, instead, needs to reset its current position to the start-of-simulation position, then I think that that would need code changes inside LinuxCNC.

It's interesting to note that, as far as I know, nobody has asked for this in the 10 years I have been working with LinuxCNC.

Perhaps a another way might be to configure an identical virtual machine config, then run that in a sim version Linuxcnc running in a virtual machine. That is what I usually do, but not at the cnc machine, I usually do it on the PC I do my CAM work from.

I am usually working with milling machines, the simulation generally in axis is as a quick test, because it was already done in a Cam system. Last Friday I had the need to program a compensated trajectory on a very large lathe that assembled with linuxcnc, and I had to remove the piece to be able to test the program until I could make the compensated trajectory work well.

It is there that I realized that you cannot test programs on Linux with Axis. I had not noticed it before.

I usually program Fanuc-Fagor-Hass, etc. and I have always simulated the programs on foot. I never realized this in Linuxcnc

It occurs to me that I could save the state of linuxcnc before when I request the simulation mode, disconnect the outside world, and when I return to the execution mode to recover the state prior to the simulation request. as if it were a call to an interruption in microcontrollers.

It would be interesting to have this type of function.

Again, the first answer can do all you are asking for.

You tell me that the solution is to have two linuxcnc working in the same machine, one as a simulator and another functional?

I think so, but the problem is they could not run both simultaneously. My suggestion of putting the sim in a virtual machine would allow them to run side by side. (Provided you could find a PC that could handle it without screwing up your latency too much.)

I can't believe your comments, but I respect them. Thanks for them, I hope linuxcnc in the future the developers can give you a more professional approach.

It is very difficult to propose improvements

It is the second time that happens to me, the first one was with the tool radius compensation, what the developers say is fine, in cnc technology it is a big mistake, the compensation loses a segment of the trajectory at the entrance, and now this. The result is always a cheap alternative, not a professional one.

Thank you all, I will continue my way

RobotMatic wrote: You tell me that the solution is to have two linuxcnc working in the same machine, one as a simulator and another functional?

Yes.
you can have many different machines on one pc. Or as many sim's as you like.
And if you want to have the machine working while you are testing something else, do it on a virtual box, as Todd mentioned.

Or if you really want to do it like Fanuc (Which I feel is a perfectly awful way). It isn't hard to disable the output to your drives in hal (including short circuiting the feed back). But you will also likely find the simular problems with that solution as Fanuc controls have. Namely you may have problems if your simulation doesn't end exactly where it started. This is why our machine operators are forbidden from using the dry run or z lock modes on those controls. On our Fanuc machines if you screw up the position with those modes the only way to undo it is to power cycle the machine.

RobotMatic wrote: It is the second time that happens to me, the first one was with the tool radius compensation, what the developers say is fine, in cnc technology it is a big mistake, the compensation loses a segment of the trajectory at the entrance, and now this. The result is always a cheap alternative, not a professional one.

I am sorry that the answers are not what you hoped for.

Fundamentally the problem is that LinuxCNC is cheap software, and you tend to get cheap solutions.

We have proposed work-arounds that might get part way to what you want, and that work _now_ rather than at some unknown point in the future when somebody decides that they fancy looking at your enhancement ideas.

Maybe you think that LinuxCNC has a whole team of developers who are allocated jobs to do by a central controlling committee? That isn't the case at all. LinuxCNC isn't anyone's job. Not even on a part-time basis.

What would your proposal look like? Would it just be the equivalent of the existing Axis preview, but running in real-time? Running the preview in actual real-time would be relatively hard, but running it line-by-line might actually be quite easy. (And less tedious for the operator)

But then you have something rather close to what (I think) already exists. I think you can click on the G-code and cursor up/down to see what each line does in the preview. (I can't guarantee that this works at the moment, as I am working away from home, and nowhere near a LinuxCNC machine, I am on my Mac laptop in a hotel). Is your requirement for a way to _force_ your operators to perform this process?