Future thoughts

On the scientific side, khmer’s future will be driven by research needs. (I think it’s clear more theory is needed, though.)

On the software engineering/community/maintenance side:

Things to discuss

  • Should we try to cater to a larger community or is it a waste of time?

    To expand:

    • one approach is to try to make khmer more broadly useful by implementing more well-rounded features (exact graph storage, for example; exact k-mer abundance tables; etc.) within the same or similar API. (This would also potentially involve more investment in community engagement - khmer days? khmer hackfests? etc?)
    • Another approach is to simply do high visibility research. Experience suggests that people will come use the software if it does highly visible things.

    Of course we’ll pursue a blended model. The question is, how blended?

  • Would more funding be useful or a hindrance? Who would pay attention to the software engineering?

  • we have a roadmap; is it still useful?

Things I’m personally not considering

Introducing support for multi-node parallelism.