Good point. I've put a note into the description to that effect.

More importantly, I've also changed with-random-source to be a procedure, like with-input-from-file. There's no reason for it to be syntax.

I've merged everything from the main SRFI repo and Arvydas's repo.

Is anyone still thinking about the spherical/elliptical distribution or the gamma distribution? Will Clinger has an implementation of the gamma function at https://github.com/scheme-requests-for-implementation/srfi-144/blob/master/srfi/144.special.scm. If you'd like to do either of these, by all means.

Still trying to understand what Brad wants.

On Sat, Jul 11, 2020 at 4:45 PM Bradley Lucier <xxxxxx@purdue.edu> wrote:

On 7/10/20 8:19 PM, John Cowan wrote:
> 3) At present, with-random-source has the form (w-r-s source proc args),
> which was meant to be a drop-in replacement for allowing the other
> procedures to take a source argument.
>
> But perhaps it should be (w-r-s source . body)? In the single-procedure
> case, that would still be only (w-r-s source (proc args)).
>
> Alternatively, and even more flexibly, it could be a procedure (w-r-s
> source thunk), in which case our single-procedure case would be (w-r-s
> source (lambda () (proc args)).
>
>
> What do people think of these alternatives?

All the other (with-... ) forms I know take a thunk.

That being said, the current version *can* take a thunk, passing a
procedure without any arguments:

(with-random-source source thunk)

I can't think of other considerations.

Brad