if someone built abstractions with the same name, is it likely that they have different behaviour?

Some of the signal objects would have to have a different behavior because of how [inlet~] works. If [inlet~] could take an optional float arg to output a constant sig (tough because it takes symbol args as well, but not impossible), then you could have abstractions which could take creation arguments *or* signals in all inlets in one abstraction implementation. Otherwise [inlet~] promotes float messages already, but it's very buggy in canvases that have an [inlet] to the left of inlet~. With the current tools it's hard to implement some of them efficiently, too.

Other problems with abstractions would be something like "gate" -- you need [initbang] for that one, which is not in vanilla (it's otherwise very easy to make an abstraction out of) -- or [spigot] could become a multi-outlet object whose only difference from [gate] is the (proper, IMO) "right inlet controls" Pd style -- this is one of the "proper behavior" problems that has come up recently; [pow~] was another, if my memory isn't shot.

One slightly different tack would be instead of trying to bring cyclone (etc.) into vanilla, to just ensure the existence of standalone control and signal objects for all the functions and operators in the expr suite -- that would be a decent start. It would at least satisfy those who want a more complete set of mathematical tools without needing to use expr; it seems that's where most of the complaints have come of late.

Matt