Peter Brier

Seems indeed related to newlib-nano. When compiling without nano specs it indeed reports a "0" on the axis.motor.error. When replacing this in [Tupfile.ini](https://github.com/odriverobotics/ODrive/blob/2922b22447f5122d558a4d702eb81d6b8570189e/Firmware/Tupfile.lua#L347): ``` -- LDFLAGS += '-mthumb -mfloat-abi=hard -specs=nosys.specs...

Hmm. Not using the newlib nano.specs causes the "w" command of ints to fail with a "not implemented" response. (e.g. "w axis.requested_state 1" --> "not implemented") so something needs to...

Yes. Pulled latest version and Checked with and without the newlib spec. 100% reproducible result. With newlib.spec: command accepted, without newlib spec "not implemented". Could this be related to the...

An would using the full newlib be wise? It takes a 2k RAM penalty (and ~7K flash). In addition: using scanf() and printf() on a resource constrained platform could be...

I confirmed the problem is indeed also related to my local arg-gcc compiler! Using the dockerbuild.sh script solves the problem of setting the axis state (without nano.specs both reading and...

Having tested various versions of the ARM compiler and with nano.spec and without I wonder why not remove the root cause of this problem? The int64 for the error flags....

Have to check. You mean arm-gcc 10-2020-q4-major for example? nano.spec will always be an issue I think for 64bit ints (limitation of library).

Dear @nebkat: Any suggestions on building from source? I tried with an official ESP IDF docker container, (version 4.3 as mentioned in the logs of the running firmware ```v0.5.4```) but...

I used an official ESP IDF docker container, (version 4.3 as mentioned in the logs of the running firmware ```v0.5.4```) but the ```button``` component source is not found (not checked...