When a bundle developer is working on their bundle, they will try it out with duffle install and probably use the docker driver to run their invocation image locally. After a failed run (or sometimes after a seemingly successful one), developers are likely to want to poke around at the state of the invocation image at the end of the run. Since the CNAB runtime injects a number of files into the image, they want to explore the actual state of the container that ran.
Currently, the docker driver always cleans up the containers, so this is not possible. A new config parameter should be added to the docker driver, and possibly documented in duffle so that it actually gets used.
When a bundle developer is working on their bundle, they will try it out with
duffle installand probably use the docker driver to run their invocation image locally. After a failed run (or sometimes after a seemingly successful one), developers are likely to want to poke around at the state of the invocation image at the end of the run. Since the CNAB runtime injects a number of files into the image, they want to explore the actual state of the container that ran.Currently, the docker driver always cleans up the containers, so this is not possible. A new config parameter should be added to the docker driver, and possibly documented in duffle so that it actually gets used.
The config parameter should probably just skip this line: https://github.com/deislabs/cnab-go/blob/master/driver/docker/docker.go#L178