Update base image

[SergioGasquez]

This would make it compatible with the latest espflash versions

[Vollbrecht]

TLTR: need python3-venv installed via apt.

debian bookworm changed some things regarding handling of virtual env, esp-idf checks if virt-env is installed and try's to install it itself via pip. Installing any python package not in a virt-env is not allowed on bookworm anymore, so it needed to be installed by apt or installed in a virt-env, put pip cant create a virt-env to install virt-env, because virt-env is not installed :D.

[Vollbrecht]

If my test finished with the new dockerfiles it would make sense to use the then new upstream container directly inside the .devcontainer.json. The one difference here currently is that it installes esp-idf globally inside the container. I thought about this a bit if it would make sense for users that generate a new template, to have an installed idf inside already in a global dir, but this would mean we would currently need two different containers per arch because of support of 4.4.6 and 5.1.1.

Furthermore this would defeat this container as a base image for no_std. I am currently checking out an minimal image for no_std all image and compare it with an all std image. My idea here is to maybe have a no_std image that really fits your use case, and we the changes for std with a COPY img into a std one. Need to play with it a bit.

For this repository we could use our new baseimage in this Dockerfile and Layer the installation of the esp-idf version ontop of it

[SergioGasquez]

For this repository we could use our new baseimage in this Dockerfile and Layer the installation of the esp-idf version ontop of it

The thing is we use a pinned nightly version on this training, to make sure everything is stable

[Vollbrecht]

For this repository we could use our new baseimage in this Dockerfile and Layer the installation of the esp-idf version ontop of it

The thing is we use a pinned nightly version on this training, to make sure everything is stable

The training assumes esp32c3 right? if we later produces a stable specialized esp32c3 image, it also would have some pinned nightly version in it, the one used on its publish day right?

[SergioGasquez]

The training assumes esp32c3 right? if we later produces a stable specialized esp32c3 image, it also would have some pinned nightly version in it, the one used on its publish day right?

If we update the training every time that we publish an image, then it would be a good thing