[ESS] proposal re: underscore, part 2

[Jack Tanner]

On 4/12/2012 9:29 AM, Rodney Sparapani wrote:
>
> It's not ideal, but since there is no = in the BUGS/JAGS language:
> my hack works well enough.

The '=' hack sounds like a clever thing to do for BUGS, although 
(define-key ess-bugs-mode-map "=" 'ess-smart-underscore) hurts my head.

> I think the idea of the BUGS language (inherited by JAGS) was to be
> S/R-ish: it's a model description rather than a programming language
> per se. BUGS is not identical to S/R, but it's almost a subset; the
> R2WinBUGS/R2OpenBUGS packages actually use the R parser on the BUGS
> model file (but I can't remember why: Gelman has blogged about this).

There are parts of BUGS that are not legal R syntax, and parts of R that 
are not legal BUGS. But there is a fair amount of similarity, also.

If you guys really want to stay ahead of the curve, Andrew Gelman and 
colleagues are developing a new modeling language called Stan, with 
another R-ish, BUGS-ish syntax (that's neither quite BUGS or R).

Here's an example of a BUGS model and how it would be written in Stan:

https://code.google.com/p/stan/wiki/Surgical

 From the development and discussion around Stan, I'm guessing it will 
be very widely used. The Stan manual, in SVN on stan.googlecode.com, is 
pretty good.