Closed megastef closed 8 years ago
I think this is a fix for #11 as well
I agree with your analysis. I definitely do not remember why the error was thrown instead of handled somewhere.
In fact, it should have been emitted at the stream level, but even that would probably have left things in a bad state.
However, we need to differentiate between two types of errors: "I can't talk to Docker" and "this container does not exist anymore, etc..". The first should be emitted as 'error'
to the result
stream, and the other can be silently ignored.
What do you think?
Released as v1.3.0.
please see my questions in #17 - there might be a better way to handle the case when a container gets killed before inspect call is finished.