how are you doing that?
I guess as proof of concept it OK
Newton constraint solver is too expensive for cloth simulation. But the math of cloth simulation are embedded in solver.
This is what I would do.
I would try to get it as good as I can possible can just as a prove of concept or prototype but never for production.
1.6 Archemidea will add a constraint particle solver that will be used for soft bodies, cloth and fluids. It will be much faster for those things.
I assume you are using ball and socket for the simulation, and if that's is truth then you can play with joint stiftness.
However I do not think it will be stable enough because the angular dof intruduce lot of instability, but you do not loss anything trying.
I had implemented cloth on comercial games before, do not worried Newton will have cloth, the reason It does not yet is because there are higher priorites in the pipe.