[epsilon-devel] Managing S-expressions in epsilon, and more

[Luca Saiu]

  Hi Matteo.
  Thanks to the hard-working Savannah guys the CVS server and the web 
interface are back up, and I'm in the process of committing all the 
changes; yet the download area won't be accessible for some time, so 
you'll have to fetch your next snapshot from CVS (go to the Savannah 
page; note that you'll need to use CVS over ssh). If you need a snapshot 
when you are in Spain don't worry, I'll mail snapshots to this list.

  Could you please send me a copy of the interpreter we wrote together 
two days ago? And an exercise for you: modify apply_ to manage primitive 
operators; it's easy if there are no errors in the file you have; but 
there could be some. Mail me a copy and I'll help you.

  The main limitation we have in writing interpreters is our current 
inability to easily write parsers. A temporary (but however very good) 
solution is using S-expressions, as Lisp does. I wrote an epsilon 
library (attached) allowing us to do this with extreme ease; try reading 
the sources. Even if there aren't many comments you should be able to 
understand everything except the scanner and the parser. The library is 
also useful for many other things, and I'm going to include it in the 
default epsilon library.
  We could even use S-expressions as a temporary concrete syntax for 
epsilon; so we would be able to write a translator from S-expressions to 
abstract epsilon syntax in *minutes*.
  For example \ x . x + 1 would be written in concrete syntax as
  (lambda x (+ x 1))
, and define y = x + 12; could become
  (define y (+ x 12))

  First tell me if you don't understand something about what I wrote; 
then we'll update an interpreter to use S-expressions, and discuss the 
datails.

  Bye,

--
Luca Saiu, maintainer of GNU epsilon
http://www.gnu.org/software/epsilon

s_expressions.tar.gz
Description: GNU Zip compressed data