Compound slabs anyone ?

Agree...

Speedikon has layered slabs, which are constructed using macros. I can see this kind of thing being easy meat for GC BIM Elements. The way Revit's Compound Walls behave I think can teach us a few things here. The individual wall layers can unlocked and extended vertically. The wall as a compound object still is manipulatable as one. It's not exploded into more objects that will be that much more tedious to manage. I think the point here is that some parent-child relationships need to be provided and managed.

Revit also has a pretty neat floor joist tool. Again, GC could be used? Revit uses a profile to define the perimeter and any openings. This could be ordinary Mstn geometry linked to GC via Element Sensors. It would be good if the 'guide' or 'parent' geometry could be used by all follow on 'smart tools' so that the outlines for the openings could be used for windows/rooflights/hatches etc made by GC,PCS,BXF etc.

What about using MCAD-style tools like Feature Solids and DDD? Do they mix well with TF solids wrt unification, re-symbolisation? Digital Project uses 2d MCAD-style profiles that are constraint solved, to extrude slabs. One very nice thing about this is the bidirectional way offsets between the gridlines and lines defining the slab edges can be manipulated. Quick, and works like you would expect, and you don't have to worry about the creation order of the elements in the sketch.

Revit has also got a 'handy' way of allowing slabs extruded in one direction to edited in a cut view that is perpendicular to the extrusion. Standard solid modeling tricks?

Levels: Correct me if I am wrong, but the way slab levels are defined is based on the datum given when the slab is first placed. This may be defined by an active ACS, recorded by the Floor Manager. But, if the floor level is changed, the slab levels do NOT update. If true, it sounds like slabs should be able to 'watch' some values or expressions like FS, DDD, XFM can?

Defining the levels in a more data-centric way would also allow a more flexible and LOD friendly result. I see a lot of designs that depicts the whole floor buildup as one element. The walls are placed on what is essentially the FFL, not the slab level. Defining the levels as parameters would save a lot of time, and allow different LOD representations.

Slab v sloped roofs: Again, Speedikon already has some powerful tools here. The 'slabs' mitre in 3d at the ridges, valleys, hips etc and this is repeated for the layers as well. Not sure, maybe the thicken solids tool can be used to produce nice 'analytical' results.

The Draw on Solid tools are great. How would this work with a layered slab?

I have been asking for composite slabs for about 3 years now, request keeps falling on deaf ears. At one point I had been answering queries as to how this would work from Bentley responding to my request but it seemed to just get forgotten about.

I had envisaged such a tool to be defined much in the same way as current composite walls but with the additional feature that you could define offsets at the edges and the height of slabs be controlled by the ACS.

If you then take this one step further and allow placement with angle as well then roofs are also covered, adding the suggestion from Sean that these could self mitre at junctions then we are very nearly at the ideal.

Seems quite simple to me although I am no programmer.  

I'm not a developer, but it sounds like a reasonable request to me.   I'd imagine one of the drawbacks up to this point is that the Slab tool was not DataGroup aware, unlike Walls.   However, those days are numbered since Building Designer will have a new Slab tool that behaves more like the other available DG tools.  So perhaps with that groundwork laid a Compound version is more practical.

It is a must since floors are a must.

We also need some smarts for the floors

When a wall goes through the floor the floor and what shoudl KNOW this is happening and have a understanding which items go through a floor and cause a hole

CMU, Ducts, PIPES, etc - not too many things

This one you mention is actually very interesting.

Right now when we want to subtract e.g. a wall from a slab (or vice versa), this is not a dynamic operation. If I move the floor, the subtraction won't update automatically. In my opinion, it should.

Things become more interesting and necessary as we are moving to implement an integrated structural model, either via ISM or via IFC. For those models it seems to be necessary that the walls meet each other. ISM wants the meeting point to be in the base of the slab, other programs like it in the centre of the plane.

If we do drawing extractions of those, we're running into a mess, because we have both slabs and walls in the same are. The same applies to massing, which gives us false volumes and areas.