Re: Mathematics skills for writing a compiler?

Rafael 'Dido' Sevilla <dido@imperium.ph>
13 Sep 2004 12:27:03 -0400

          From comp.compilers

Related articles
Mathematics skills for writing a compiler? lucky7456969@hotmail.com (2004-09-08)
Re: Mathematics skills for writing a compiler? sandra@frogsonice.com (Sandra Loosemore) (2004-09-13)
Re: Mathematics skills for writing a compiler? dido@imperium.ph (Rafael 'Dido' Sevilla) (2004-09-13)
Re: Mathematics skills for writing a compiler? torbenm@diku.dk (2004-09-13)
Re: Mathematics skills for writing a compiler? dmclean@stsci.edu (Donald F. McLean) (2004-09-13)
Re: Mathematics skills for writing a compiler? napi@axiomsol.com (2004-09-14)
Re: Mathematics skills for writing a compiler? vidar@hokstad.name (2004-09-14)
Re: Mathematics skills for writing a compiler? joe@burgershack.com (Randy) (2004-09-21)
Re: Mathematics skills for writing a compiler? kamalp@acm.org (2004-09-21)
[3 later articles]
| List of all articles for this month |

From: Rafael 'Dido' Sevilla <dido@imperium.ph>
Newsgroups: comp.compilers
Date: 13 Sep 2004 12:27:03 -0400
Organization: Compilers Central
References: 04-09-063
Keywords: practice
Posted-Date: 13 Sep 2004 12:27:03 EDT

On Wed, Sep 08, 2004 at 12:04:30PM -0400, Jack wrote:
> What mathematical skills do I need in order to build an "average" compiler?
> such as numerical methods, CFG, DFS.... etc


To write a compiler, the main thing you need is some background in
formal language and automata theory. All three levels of the Chomsky
Hierarchy are used in the construction of a compiler: most lexical
analyzers are based on regular languages, most parsing is done using
deterministic subsets of context-free languages, and well, all target
architectures are essentially linear bounded automata that can be
idealized as Turing machines. Naturally algorithms and data structures
are a necessity, and graph theory is certainly useful, in order to
swallow some of the techniques for code generation and optimization.
Numerical methods are not needed unless you're planning on designing the
successor to FORTRAN, a dialect of Matlab, or some other language that
specifically has numerical analysis as its problem domain.


--
dido
Te capiam, cuniculus sceleste!



Post a followup to this message

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