Johan Mabille

I think we should wait before merging this. I have opened a [JEP](https://github.com/jupyter/enhancement-proposals/pull/82) to move the `kernel_info_request` to the control channel to fix https://github.com/jupyterlab/jupyterlab/issues/10174, and the jupyter server expects both...

For the first one, you are comparing different things a simple statement in the case of cling, vs a statement nested in an anonymous namespace in the case of xeus...

Hi, I have started to write a C++ implementation of a client (or, more precisely, of a framework for writing a client) for Jupyter, but it's far from being usable....

Hi, Indeed I didn't have enough time to continue this side-project, I wil push what I have locally and open a bunch of issues of how I see further developments.

The invocation of cling meta-processor is done [here](https://github.com/jupyter-xeus/xeus-cling/blob/master/src/xinterpreter.cpp#L112). If I remember correctly, vanilla cling does not use the meta-processor, so it might be worth checking the differences between these two...

It should work if you define the function in a dedicated cell.

Yes, I've been working on the conda recipe since yesterday.

Can you give the output of the following commands? ```bash mamba list |grep xeus mamba list |grep zeromq otool -L xcpp ls -al $CONDA_PREFIX/lib |grep xeus ```

I don't know all the packages depending on ZeroMQ, but at least all the Jupyter kernels, the jupyter server (required for Jupyter Lab and the notebooks). Therefore you would need...

A better option would be to add this message as a part of the `xkernel::start` method: ```cpp void start(const std::string& kernel_name, const std::string& config_file_name); void start(const std::string& kernel_name); ``` That...