Multiple output file references

We make an implicit assumption all of the above that the output of a pipeline step is a single file, or a single directory or a set of files that relate to each other. What we don’t cover yet is a component that outputs both the results of analysis and a report, for instance. The reason for this is not technical but rather that it breaks the logical flow of a pipeline definition.

We currently output 1 type of output from a module, and that allows us to easily chain modules. A module that would output 2 types of data would essentially be a fork in the pipeline process. That means that either the next component knows to expect these two outputs as inputs or we have to explicitly deal with the two branches of the fork. But then we can not longer write pipelines like:

step1 | step2 | step3

There is a simple workaround for this kind of situation: Make two components/modules that each output either of the outputs. This fits in the API of a modules and those can again be chained easily. Having said that, we are thinking of ways to allow the output of multiple output files because this workaround is not efficient at all.