4b. Though after all 40 slides I don't see how we have an actual virtual filesystem that other programs will believe. Trust in magic.. I want to see the output from "ls" though, I'm not quite convinced yet that will work. But then again I don't need to run that yet, and could maybe make a module to do run such an external program if there was a problem.

Like I already said, a zip-file is a filesystem. I don't know of any "ls" program that deals with zipfiles, but then you wouldn't want to use "ls" to manipulate zip files. See Accessing packed files in PAR::Tutorial

5. I notice the multiarch switch. Too much to ask for me to be able to develop it all in linux and deployed in windows?

I develop stuff on windows and deploy it on linux (and vice versa) all the time :-) All the multiarch switch does it sets up all the modules in the archive in an architecture/platform specific directory( like foo.par\MSWin32-x86-multi-thread\), so you can have a single par archive usable on many platforms (i don't see the appeal, it just bloats size).

You still need to build a binary for each of your target platforms(par doesn't support that yet).

PAR also doesn't support creating such a multiarch .par (par cannot leverage ppm...).

What I just described is sort of a wishlist item (I recall it being discussed somewhere, and I believe its somewhere on the TODO but it's a long ways off).