#6 Merged
  1. Matt Turk

This is a YTEP that describes periodicity changes.

I would like feedback from implementors of the various objects described as well as from those who rely heavily on those objects. However, after writing this, I think the problem is simpler than I initially thought so I would like to make progress on it in the near future.

One open question is whether or not this should be a part of 2.5 or only in the 3.0 series.

I'll open this up for discussion on the mailing list as well, but I would like signoff from @Stephen Skory , @Britton Smith and @Nathan Goldbaum (all of whom have written periodic functions, reported bugs or used periodic functions extensively). @John ZuHone has also written some for the FLASH particle reader.

UPDATE 1: Changed periodicity to be tuples.

Comments (4)

  1. Nathan Goldbaum

    Hi Matt, thanks for thinking about this. I had marked myself for the bug about periodicity in the Radius field with the intention of working on it when I had some free time. Like you mentioned, I think going through and doing this won't be so bad.

    However, one thing you don't mention below which we do need to consider in concert with the frontend maintainers is that there needs to be a code-dependent way if checking for periodicity. I know how to do this for enzo and FLASH but would need to dig through the documentation of other public codes to implement this. Private codes with no public docs will probably require a bit of work from the frontend maintainer.

    I'd still be happy to get started on implementing this using the YTEP as a guide. Let me know if you had another plan in mind.

  2. Matt Turk author

    Hi Nathan, thanks for the reply! You're quite right, I completely missed the issue of determining how to identify whether or not a simulation should be regarded as periodic. A continuing conversation with code frontend maintainers might be outside the realm of this YTEP, but one thing we can and should do is enumerate in the YTEP how we mark the periodicity flag for each code frontend, and adding to that as time goes on. Does that sound okay?

    As for getting started on implementing it, that would be really awesome if you would be up for doing it! If you are ready to try your hand at it, I think this can probably be accepted and you should feel free to start. I'm also happy to help if you'd like it, a bit down the road.

    1. Nathan Goldbaum

      That sounds fine. As a first pass I'll just set periodic to True when I can't determine how to see if a simulation is periodic or not, this will presumably recover the current behavior.