Build and deployment pipelines for Azure Devops - problems with PBI DATAFLOWS

Anonymous 

Thanks for the pointers to those.

I'm still having trouble at a conceptual level understanding why dataflows are so tricky to deploy, when datasets are not.  They have similar features, such as parameterization, and pre-environment connections, and gateways.  Despite some similarities, the datasets certainly seem to have the advantage.  They have the required API's to support automated deployment, but dataflows are extremely inconvenient to deploy.  

Can you please explain what factors went into the design, that causes dataflows to be so unfriendly to work with, as compared to datasets?