From mboxrd@z Thu Jan  1 00:00:00 1970
X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me
X-Spam-Level: 
X-Spam-Status: No, score=-1.3 required=5.0 tests=BAYES_00,INVALID_MSGID
	autolearn=no autolearn_force=no version=3.4.4
X-Google-Language: ENGLISH,ASCII-7-bit
X-Google-Thread: 103376,37680a99b5e22b2b
X-Google-Attributes: gid103376,public
From: dewar@merv.cs.nyu.edu (Robert Dewar)
Subject: Re: Shared Generic Instance Code
Date: 1997/04/04
Message-ID: <dewar.860206721@merv>#1/1
X-Deja-AN: 230825781
References: <5hrkhkINN9ip@snoopy.cis.ohio-state.edu>
  <1997Apr1.201631.28634@ocsystems.com> <dewar.859987468@merv>
  <sO6r4AADiBRzEwoj@marnhull.demon.co.uk> <dewar.860114590@merv>
  <5i23k6$hkq@mulga.cs.mu.OZ.AU> <dewar.860160858@merv>
 <zoIMRAAI9XRzEwoa@marnhull.demon.co.uk>
Organization: New York University
Newsgroups: comp.lang.ada
Date: 1997-04-04T00:00:00+00:00
List-Id: <comp.lang.ada>


Bill Keen said

<<My post was prompted by thought that if the generic actual parameters
can be determined at compile time then the code for a generic instance
can be a straight inline expansion of the body. But if not, then the
generated code must contain some mechanism for choice. I thought this
constituded code sharing, though I see that the subject is much wider.
But my curiosity is aroused. How do compilers (does GNAT) deal with this
distinction?>>

GNAT uses macro expansion for all generic specs and bodies. There is nothing
about the compilation model or structure of GNAT that dictates this choice,
it is a deliberate design choice that favors efficiency over space. Eventually
it would be nice to implement generic sharing at least on an optional basis.