This project has moved and is read-only. For the latest updates, please go here.


Restructure namespaces for data structures


Proposal: Re-organize data structures into a single location for consistency, discoverability, and future interoperability (e.g. between FEM and MC solvers)
  • Root folder and namespace of all serializable data structures should be Vts.DataStructures
  • Sub-namespaces would include Vts.DataStructures.MonteCarlo.SourceInputs/DetectorInputs/etc, Vts.DataStructures.Fem.SourceInputs/etc
Closed May 5 at 8:35 PM by hayakawa
Not needed anymore since no plan to update FEM.


hayakawa wrote Feb 27, 2012 at 7:23 PM

If FEM and MC solvers will possibly share sources, then wouldn't SourceInputs be at Vts.DataStructures.SourceInputs level? And if so, are we going to have to rename when we find that a data structure can be used by more than one solver?

dcuccia wrote Feb 27, 2012 at 8:44 PM

Yes, and yes, we'll have to rename at that point. Right now we can't prevent having two separate locations, since they are currently not shared structures. This would be laying the groundwork for consistency.

There are a few things that pull in the opposite direction of this proposal (e.g. locating inputs and sources together). I have seen this as I've been working on VTS extensibility (starting with the detectors), and this raises a few more questions regarding locations of files. For example, if the "end user" should be able to define a detector (and detector inputs) in a separate assembly, and have it "just work" with the existing infrastructure, then there's a strong push to have all of that easily defined in either 1) one file, and/or 2) with the users own namespaces. Clearly, this "story" doesn't jive with the internal structural conventions. Perhaps this doesn't matter. In any case, I will bring this up at Wednesday's meeting to present my work so far on this topic.