Re: Back End Generators

Peter.Damron@Eng.Sun.COM (Peter C. Damron) writes:
>Graham & Glanville did the first (tree) parser for code generation.
>They did not do dynamic programming for parse selection.
>The LR parsing technology they used could not accommodate this.
>
>The first tree pattern matching (really tree parsing) based code
>generators that used dynamic programming were done by Aho, Ganapathi, &
>Tjiang at Stanford/Bell Labs, by Henry & myself at Univ. of Washington,
>and by Hatcher & Christopher at ??? (see bibliography below). Also, Chase
>and Pelegri-Llopart were doing work in this area at about that time
>(1985-1986).


Vaughan Pratt's 'Lingol' system for natural language used a
combination top-down, bottom-up strategy, and used inherently
ambiguous context free grammars and a sort of 'dynamic programming'
for English. I realize that it isn't 'code generation' for computer
instructions, although Vaughan did use this for parsing commands for a
Winograd-shrdlu-type robot, and this work was done in 1971-75. I will
be making a Common Lisp version of Lingol available on my ftp
directory in a few weeks (I worked on Lingol for a while as a graduate
student).




            Henry Baker
            Read ftp.netcom.com:/pub/hbaker/README for info on ftp-able papers.
--

Return to the comp.compilers page.
Search the comp.compilers archives again.