ColorCNC Colorlight 5A-75E/5A-75B as FPGA controller board

[/code][/code]
[/code] [/code][/code]
[/code] [/code][/code]

Thanks for your finding. I modified the documentation, the line should read:
NOTE: I used xvel_cmd as signal name, as this is the velocity command. It is an unique signal, if you had appointed an existing signal to this line, please rename it.

Okay Thank you. And for the other axes I take yvel and zvel?

juergenmoser wrote:

Okay Thank you. And for the other axes I take yvel and zvel?

Would recommend yvel_cmd and zvel_cmd, but any name will do as long as they are unique.

@deroj: I changed a small bit in my modified halcompile to show the errors (added print("EXTRA_CFLAGS += -Wall", file=f) to line 1117). Now I can reproduce the warnings you got. Going to solve these warnings, as I believe in clean code is better code. But, how did you get those warnings on your system, did you change halcompile as well or have a certain setting somewhere? Otherwise this is also a difference between the 2.8-family and 2.9.

Edit:
I solved now all compiler warnings in the code in the branch 7-resolve-compiler-warnings. I will leave the option -Wall as default to see any new warnings. In my opinion the driver should compile without any warning.

I also started on testing why the code fails to link on LinuxCNC 2.9. Created a new development container with Debian Bullseye and LinuxCNC 2.9.0-pre1, et voila, I can reproduce the error:
Now have to solve this one...

@TOLP2 Not, as I know...I just installed Linux Mint 21, then the debs with preempt kernel and linuxcnc, and then followed your python install instructions

The warnings are solved now. The mostly stem from difference between 32-bit systems and 64bit and the formatting of numbers. Now I use inttypes to format the numbers, which solves the warnings. There were some other warnings as well, those have been solved at the same time.

For compiling it in LinuxCNC 2.9: something has changed in the linking; there is a notion of it in the CHANGELOG. However, I cannot find a way to successfully link the external library after this change. I decided to solve this by changing libjson-c to cJSON. The benefit of using cJSON is that this is a small library which can ben inlined in the driver. This requires a re-write of the driver on the configuration part. I rewrote the driver litexcnc_eth as a prove of concept and it works really well. 

The major work is now in rewriting the modules and the main driver.

Edit:
For the brave: I've committed the re-written code to GitLab. Didn't have the chance to test the code fully, but preliminary tests show that all pins are correctly created from the JSON. Took also the chance to refactor the initialization for the modules to get them all in the same format. The one thing I'm not confident about yet is the calculation of the CRC fingerprint.

So, if you are feeling adventurous, you can try it. Think I can do some testing myself during the weekend.