[comp.compilers] compiler compilers

tarvydas@turing.toronto.edu (Paul Tarvydas) (11/10/87)

I don't have experience with this one, since it's fairly fresh, but
Cordy's PhD thesis describes a beautiful algorithm for generating
coders.  I do have experience with Holt's and Cordy's (et al) earlier
stuff (Concurrent Euclid written in S/SL) and I think that it's
wonderful, especially the concept of 'data descriptors'.  The source
for Concurrent Euclid (the language) and S/SL (the tool) are
available.  S/SL may not be exactly what you're looking for - it
forces abstraction, thus allowing you to stay sane while building
extremely complex coders, but you have to build the coders manually
(unless you take the original ConEuc coder and butcher it as a
starting point).

Cordy's thesis is a must read:
	"An Orthogonal Model for Code Generation"
	James R. Cordy
	Technical Report CSRI-177
	January 1986
	Computer Systems Research Institute
	University of Toronto
	Toronto, Canada
	M5S 1A1

A paper about data descriptors was published in TOPLAS:
	"Data Descriptors: A Compile-Time Model of Data and Addressing"
	R.C.Holt
	ACM Transactions on Programming Languages and Systems
	July 1987, Vol 9, Number 3
	pages 367-389

Paul Tarvydas
Tarvydas-Sanford Controls Inc.
tarvydas@csri.toronto.edu
--
Send compilers articles to ima!compilers or, in a pinch, to tower@bu-cs.bu.edu
Plausible paths are { ihnp4 | decvax | cbosgd | harvard | yale | cca}!ima
Please send responses to the originator of the message -- I cannot forward
mail accidentally sent back to compilers.  Meta-mail to ima!compilers-request