Result --> SingleResult

[loriab]

So we've got

• OptimizationInput/Optimization
• ResultInput/Result

And in another project, I've got the potential for

• FindifResultInput/FindifResult
• CBSResultInput/CBSResult

etc. The optimization pair doesn't particularly signal "schema" to me. And it might be convenient to distinguish between the set of *ResultInput/*Result pairs and the particular energy/gradient/hessian calc that ResultInput/Result now represents.

It is suggested that the two first items above become the below so that the * of *ResultInput/*Result is never the empty string.

• OptimizationResultInput/OptimizationResult
• SingleResultInput/SingleResult (AnalyticResult?)

Please weigh in on this change, even if it's just an up/down emoji response.

[mattwelborn]

@dgasmith Does this proposal interact at all with plans to make optimization a service?

[dgasmith]

@mattwelborn No, doesn't effect those plans.

Another option is to do:

• OptimizationInput / OptimizationResult
• SingleInput / SingleResult
• FindifInput/FindifResult
• CBSInput/CBSResult

Since these names are going to be typed many times I would be hesitant to make them too long where Resultinput seems a bit duplicitous.

It seems that the main blocker is that Result / ResultInput is odd name. Is there a better name for the atomic unit of a component framework?

Worth poking @dotsdl @Lnaden @jchodera @bennybp.

[jchodera]

I'm with @dgasmith here: It's clearer to use XInput/XResult, or perhaps even XInput/XOutput. ResultInput is syntactically painful.

[loriab]

I've thought ResultInput a pleasing quirk because you could search for CBSResult and hit both input and output stages. But sure, I don't mind XInput/XResult. It's searchable and predictable (latter not the case at present).

So any votes among the below? I'm plenty happy with SingleResult. It's distinctive but with a vague enough definition to not feel wrong stretching it.

• SingleResult
• AnalyticResult
• PrimaryResult
• AtomicResult

[dgasmith]

QuantumResult for the word play, both a quantum unit and a quantum chemistry one. I will see myself out...

[dgasmith]

QuantaResult?

[loriab]

Going with SingleResult/SingleInput and OptimizationResult/OptimizationInput? I get the feeling ResultInput is not well loved. My second-favorite for SingleResult is AtomicResult, which also makes a nice play on words, so long as it's not mistaken for single-atom result.

To drum up opinions, i'll query on a few more alternatives:

• CBSResult (currently) or CompositeResult
• FinDifResult (currently) or FDMResult or FiniteDifferenceResult or StencilResult
• NBodyResult (currently) or MBEResult or ManyBodyResult or FragmentResult

[dotsdl]

I'm in favor of syntactic simplicity where it can be found, so something like XInput / XResult as a convention is something I can get behind. As a relative newcomer to the project wrapping his head around a rapidly-evolving set of tooling, it helps if the names of things give a sense for their structural purpose and place within the landscape.

ResultInput, for example, feels like a sign with one label pointing in two directions. It might get you to where you want to go, but you might also scratch your head every time before you get there.

[dgasmith]

@dotsdl Any thoughts on AtomicInput/AtomicOutput?

@mattwelborn Can you weigh in here?

@leeping @j-wags Any thoughts here? I think your comments here would be quite valuable as you are not too deep into the ecosystem.

[dgasmith]

Last round of pinging, I am up for making the change this release with a deprecation warning.

@bennybp @twindus @lothian

[leeping]

Thanks for asking; after reading this discussion I think XInput / XResult or XInput/XOutput is the more intuitive choice. :)

[loriab]

I strongly favor XInput/XResult over XInput/XOutput.
I'm for

• AtomicResult
• CBSResult
• FDMResult
• MBEResult
• OptimizationResult

with second choice

• SingleResult
• CompositeResult
• StencilResult
• ManyBodyResult
• OptimizationResult

(or any combination thereof)

[dgasmith]

It seems XInput/XResult has the day overall, nobody has yet to object to Atomic either.

For the others:

• CompositeResult
• FiniteDifferenceResult
• ManyBodyResult

FDMResult and MBEResult are hard to understand. CBSResult would be an oddity, but is a well known literature description.

[dgasmith]

Closed by #167, thank you everyone for the comments and feedback.