Proposal (Procedural) - Let's Decide Phases

[TotalTechGeek]

Background

@json-logic/tc @json-logic/core

Hey guys, our proposal lifecycle is documented here.

Since I was trying to organize the initiative, I kinda made some decisions unilaterally early on to try to get the TC going & establish a process.

Because of this, I tried to keep certain parts vague to allow us to decide some of the specifics together.

At this point, we have a handful of proposals in phase 1 status, which I consider to be a good thing!

We've made some decent progress on defining improvements & changes to JSON Logic.

However, it might be time for us to try to decide on:

• What is the threshold for Phase 1 -> Phase 2?
• What is the threshold for Phase 2 -> Phase 3?

And clarify on what Phase 1 / Phase 2 really seek to accomplish.

Proposal

Phase 1

I believe Phase 1 should be used to validate the viability of a proposal and difficulty of implementation.

In Phase 1, multiple projects should try to implement the proposal. If there is significant difficulty encountered, we can pull the proposal back for further evaluation.

I believe the threshold for progressing should require either: 5 Projects or 3 Implementers, across different ecosystems (as in different languages essentially).

I realize this convention is a bit odd: Though the purpose of it is to try to encourage independent evaluation / breadth across multiple ecosystems.

This is intended to avoid a situation where someone could try to cram a proposal through by adopting 3 projects in similar ecosystems and slapping the changes into all of them.

Phase 2

I believe Phase 2 should be a larger roll-out of the proposal across the ecosystem, but indicates that the proposal is not widely supported.

I believe the threshold to progress should be two thirds of the projects represented by this initiative, rounded up. Currently, 25 Projects are represented by the initiative, so 17 projects would need to be compatible with the proposal for it to enter Phase 3.

Once the threshold is met, it should stay in Phase 2 unless a reasonable concern comes up that should cause it to fall back to Phase 0. (I'm noting this to avoid a situation where a project joins the initiative and someone asks "Should this be pulled back to Phase 1?")

tl;dr

• Phase 1 -> Phase 2: 5 Projects or 3 Implementers. Each implementation needs to be in a different ecosystem.
• Phase 2 -> Phase 3: 2/3 Projects represented by this org rounded up.

[TotalTechGeek]

We also might need to identify volunteers / projects identified for Phase 1.

Currently, datalogic-rs and json-logic-engine serve as the primary guinea pigs for the proposals.

If we cannot find another implementer with spare time, I may try to contribute to another project or two to push us towards the 5 project threshold (hence the escape hatch, but I wanted to try to ensure some ecosystem diversity).

[codetiger]

Am ready to pick up another project other than my rust based implementation datalogic-rs.

[toddbaert]

Just for clarification, am I to understand that when you say "project" you mean a specific implementation of the normative parts of the specification in a particular language, and when you say "implementer" you mean a person in charge of the development of the above?

Assuming I'm understanding correctly:

I think ⅔ might be somewhat high. IMO, ½ would likely be enough of a sample size to establish feasibility. I'm fine with sticking to ⅔, but I think it might slow us down while only providing marginal additional assurance.

[TotalTechGeek]

"project" you mean a specific implementation

Yes; a specific implementation. Like, json-logic-engine vs datalogic-rs.

when you say "implementer" you mean a person in charge of the development of the above

Correct.

I'm okay with 1/2 if other folks are 😄.