This makes it hard to define migrations that need to discover the IPs dynamically (eg. when cassandra's running in a kubernetes cluster).
A possible solution would be to create a reader that lets client code supply a function that can fill in the data necessary. Suppose that reader's tag is #foo. Then the configuration data might look like:
The cassandra configuration requires there be a vector of hosts:
But the readers for the tagged literals
#env
and#envf
can only return strings.This makes it hard to define migrations that need to discover the IPs dynamically (eg. when cassandra's running in a kubernetes cluster).
A possible solution would be to create a reader that lets client code supply a function that can fill in the data necessary. Suppose that reader's tag is
#foo
. Then the configuration data might look like:The joplin reader function would then look something like:
Of course,
#foo
's not the best tag for this reader, but I can't think of a great one at the moment.Thoughts?