Re: [Caml-list] Re: Where's my non-classical shared memory concurrency technology?

[Martin Berger <M.Berger@doc.ic.ac.uk>]

> In short, with monadic threads, you can safely invoke non-concurrent 
> code from concurrent code.  (The inverse
> can be dangerous - but you usually don't do this anyway since you will 
> end up optaining an 'a Lwt.t).

In my experience that rarely works, in the sense of scaling to code of
significant size.

Such code must be written with the concurrency mechanism at the forefront
from the start.

I don't agree that event based mechanisms prevent syncronisation bugs
or consistency bugs as you call them, but the issue has already been
taken up by others in this thread, so I shall keep this message short.

Martin