Mixed compilation

dalamb@qucis.queensu.ca (David Lamb)
Fri, 31 Jan 1992 14:30:13 GMT

          From comp.compilers

Related articles
Mixed compilation dalamb@qucis.queensu.ca (1992-01-31)
Re: Mixed compilation goldberg@iitmax.iit.edu (Bob Goldberg) (1992-01-31)
mixed compilation -> partial evaluation dalamb@qucis.queensu.ca (1992-02-01)
Re: Mixed compilation blenko-tom@CS.YALE.EDU (1992-02-01)
partial evaluation debray@cs.arizona.edu (Saumya K. Debray) (1992-02-04)
Re: Mixed compilation daniel@quilty.Stanford.EDU (1992-02-06)
| List of all articles for this month |

Newsgroups: comp.compilers
From: dalamb@qucis.queensu.ca (David Lamb)
Keywords: optimize, question
Organization: Computing & Information Science, Queen's University at Kingston
Date: Fri, 31 Jan 1992 14:30:13 GMT

Does anyone have any references for A.P.Ershov's work in the late 60's or
early 70's on "mixed compilation", or on any modern work that fits the
following description:


        Take a code fragment (eg procedure) and specify values for *some* of
        the input variables (parameters, tables, etc.) that it reads (and I
        presume doesn't write). Wherever the code refers to these
        "constants", perform all the usual constant folding, constant
        propagation, and so on, generating code only for manipulations that
        can't be done at "compile time". The example I especially remember
        (from a verbal description about 12 years ago) was taking a
        table-driven parser skeleton and a particular parse table, grinding
        through the "mixed compilation" system, and winding up with a
        "hard-coded" parser that was much faster (albeit larger) than the
        original.


I don't care about this specific application; I'd just like to find any
papers giving a clearer description of the technique. I've looked through
all my compiler books and one online compiler bibliography, but can't find
any references.
--


Post a followup to this message

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