AE: Group Layers without Pre-Composing

Grouping soemthing visually is not a problem, but apparently one would have to define what groups are supposed to do in terms of functionality with things how they affect render order, transform operations and a gazillion scenarios with combinations of layer switches, effects, time manipulation and so on that regular users who never have dug so deep aren't even aware of. This debate has been going on for literally as long as AE exists and will likely go on fo just as long in the future. As an oldschool guy I'm already loathe of the many things added in recent years like the pointless compositing options which only muddy up a stringent workflow, so I'm definitely pretty much opposed to any major changes and introducing layr groups in teh timeline. For me it really comes back to what I've been saying for forever: People need to learn to see pre-compositions and other stuff as valuable tools, not their enemy they need to avoid or look for ways around. That's pretty much all there is to say on the matter.

Mylenium

GM FoldLayers looks pretty neat.

When I've seen this discussed it's been that people effectively want to be able to 'twirl open' a precomp and see it's layers visible in the current comp. This is very different from FoldLayers (which just appears to use AE's Shy Layers system to show/hide layers) in that you'd be able to affect the group of layers by altering that main layer (i.e. the way a precomp becomes a single layer in the current comp). But there are potential  issues with this, notably that you could keyframe a Time Remap onto that comp layer to change the timings of the contents, so if you were to twirl open that precomp the keyframes in it wouldn't be at times that represented the remapping occuring on the 'group layer' which would make for some potentially unintuitive editing behaviour.

I can't say for sure but I perhaps the reason we haven't seen it is because there's this expectation for what it would need to be able to do that would be considerably more complex to implement. Why do it in this more basic way when third-parties are already perfectly able to implement it themselves?