Hello Metaflow team! I am a strong advocate for the conda ecosystem, so it felt natural to me trying to leverage the

@conda

decorator offered by Metaflow. I have two questions there that I will decouple in two separate posts. First conda question: I am leveraging conda to manage my compute environment. This is done independently from Metaflow, in a docker image that serves as base for my workflow of workflows. And this environment covers all basic dependencies for my flows. There is one step from one flow for which I need to recreate a dedicated environment based on a conda env file. However I just found out that the conda env bootstrapping from Metaflow would not leverage my conda installation, so it forces me to specify the dependencies I already manage under the hood. Is there a way to tell Metaflow about using an existing conda installation for bootstrapping? Like the base environment (or any env to be used by default) would be available already and Metaflow would only have to create the step env on the fly?

@elegant-lamp-40213 in that scenario, you can skip the @conda decorator and just execute your flow using the Python interpreter of the conda environment you have already set up

That is the default behavior for my flows. In the case of this specific step, I need a different environment. I can create one independently from Metaflow, however it is less elegant, and there is a technical overhead to implement boilerplate code for creating the env and cleaning it afterwards, and to maintain the boilerplate code.