Should we prefer SDOs that handle multiple productions by branching on parse node presence?

[gibson042]

The spec currently has separate algorithms for Evaluation of ObjectLiteral : `{` `}` vs. ObjectLiteral : `{` PropertyDefinitionList `,`? `}`:

1. Return OrdinaryObjectCreate(%Object.prototype%).

vs.

1. Let _obj_ be OrdinaryObjectCreate(%Object.prototype%).
1. Perform ? PropertyDefinitionEvaluation of |PropertyDefinitionList| with argument _obj_.
1. Return _obj_.

But as noted at #3938 (comment) , I would prefer to handle them in one:

1. Let _obj_ be OrdinaryObjectCreate(%Object.prototype%).
1. If |PropertyDefinitionList| is present, perform ? PropertyDefinitionEvaluation of |PropertyDefinitionList| with argument _obj_.
1. Return _obj_.

[michaelficarra]

I feel like this is something we're going to have to address on a case by case basis.

[gibson042]

Yes, probably. We can start with the case above, which is about as simple as it gets (adding an If $optionalParseNode is present condition to the middle of a three-step algorithm and removing its single-step sibling).

[gibson042]

Editors were mostly neutral on this, but at least one implementer favors consolidation. We don't necessarily want to capture a formal convention, but are open to seeing where this leads.