Re: Object Oriented Compiler Design Problem

mikee@cetasoft.cog (Mike Enright)
13 Sep 1998 22:29:24 -0400

          From comp.compilers

Related articles
Object Oriented Compiler Design Problem mayurnaik@my-dejanews.com (1998-09-05)
Re: Object Oriented Compiler Design Problem qjackson@wave.home.com (Quinn Tyler Jackson) (1998-09-13)
Re: Object Oriented Compiler Design Problem qjackson@wave.home.com (Quinn Tyler Jackson) (1998-09-13)
Re: Object Oriented Compiler Design Problem dwight@pentasoft.com (1998-09-13)
Object Oriented Compiler Design Problem dboucher@locus.ca (Dominique Boucher) (1998-09-13)
Re: Object Oriented Compiler Design Problem brueni@ipass.net (Dennis Brueni) (1998-09-13)
Re: Object Oriented Compiler Design Problem mikee@cetasoft.cog (1998-09-13)
Re: Object Oriented Compiler Design Problem jucie@uol.com.br (Juciê Dias Andrade) (1998-09-13)
Re: Object Oriented Compiler Design Problem danwang+news@cs.princeton.edu (Daniel C. Wang) (1998-09-18)
| List of all articles for this month |
From: mikee@cetasoft.cog (Mike Enright)
Newsgroups: comp.compilers
Date: 13 Sep 1998 22:29:24 -0400
Organization: CetaSoft (com not cog)
References: 98-09-019
Keywords: OOP, design

On 5 Sep 1998 01:20:58 -0400, mayurnaik@my-dejanews.com wrote:
[OO compiler stuff snipped]
>
>The function symbol* add(symbol*, symbol*) cannot be made a virtual function
>of class symbol, since it is a friend function. But, the moment it is a
>friend function, it does not know whether the symbols to be added are
>Constants or Variables


This is a classic case of the need for double-dispatch (choosing a
function based on two types instead of one). A modern approach would
be based on the Visitor pattern. I would take this up at the
comp.object newsgroup or on the OTUG mailing list (go to www.oma.com
to sign up for that). (I'm afraid I don't have much more to offer on
the subject; I'm posting to ensure that these ideas are at least
mentioned.)


>[I suspect a union with a type field is it. RTTI in this case would amount to
>the same thing. -John]


I'm certain that RTTI is not necessary.
--


Post a followup to this message

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