On Fri, Jun 26, 2020 at 1:52 PM John Cowan <xxxxxx@ccil.org> wrote:

Great, thanks!  I think we can go to last call after that, but this is a SRFI that's inherently prone to scope creep. 

:-)

 what distributions they might actually use, speak up now. 

I do not currently use the gamma distribution, but would recommend it for inclusion, based on the following:

(quoting wikipedia): The chi-square distribution is a special case of the gamma distribution and is one of the most widely used probability distributions in inferential statistics, notably in hypothesis testing and in construction of confidence intervals

Supporting something that is widely used would seem to make it a good candidate for inclusion. (I was attracted to gamma because it is a building-block for the dirichlet distribution, which is quite easy, almost trivial, if one has the gamma distribution.)

Another important consideration is that implementing the gamma is actually kind-of hard: a look at the wikipedia page for it does provide rather detailed implementation notes. As a coder, one is always faced with a decision to roll-your-own vs. use-someone-elses.  Anything that is obvious, easy, requires no thinking is a candidate for roll-your-own. Anything that is hard or inconvenient should be a prime candidate for a srfi.

 

It would be good if you provided code.

When I was 9 years old, someone remarked that I do good things when the spirit moves me. I think that was supposed to be a remark as to the unpredictability of when I might volunteer to do something. I'm still unpredictable in that way.

--linas