Re: Abstract Interpretation vs DFA in Clang and GCC

torbenm@diku.dk (Torben Ęgidius Mogensen)
Tue, 15 Feb 2011 15:47:16 +0100

          From comp.compilers
Related articles
Abstract Interpretation vs DFA in Clang and GCC ilikequoting@katamail.com (Jack Smith) (2011-01-10)
Re: Abstract Interpretation vs DFA in Clang and GCC torbenm@diku.dk (2011-01-17)
Re: Abstract Interpretation vs DFA in Clang and GCC ilikequoting@katamail.com (Jack Smith) (2011-02-09)
Re: Abstract Interpretation vs DFA in Clang and GCC torbenm@diku.dk (2011-02-15)
| List of all articles for this month |
From: torbenm@diku.dk (Torben Ęgidius Mogensen)
Newsgroups: comp.compilers
Date: Tue, 15 Feb 2011 15:47:16 +0100
Organization: SunSITE.dk - Supporting Open source
References: 11-01-034 11-01-069 11-02-010
Keywords: analysis, optimize
Posted-Date: 17 Feb 2011 01:28:48 EST

Jack Smith <ilikequoting@katamail.com> writes:


> On 17 Gen, 12:14, torb...@diku.dk (Torben Fgidius Mogensen) wrote:
>> On the other hand, DFA can do analyses that are difficult to
>> express as AI, such as liveness. To do liveness analysis with AI you
>> need a continuation-passing semantics.


> could you go more in deep about liveness analysis?
> why would i need a continuation-passing style semantic? have you got
> some paper about that?


Liveness of a variable is not a property of the possible values of the
variable, so an abstract interpretation that abstract sets of values can
not directly express liveness.


A continuation is a functional representation of the future computation
from the present point until termination. Since liveness is a property
of the future computation, you can express liveness as an abstraction
over the set of possible continuations.


Torben


Post a followup to this message

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