Path: utzoo!mnetor!uunet!seismo!sundc!pitstop!sun!aeras!dgcad!roger
From: roger@dgcad.UUCP (Roger Scott)
Newsgroups: comp.lang.c++
Subject: Re: Cfront output (partly) explained
Message-ID: <97@dgcad.UUCP>
Date: 29 Feb 88 19:31:53 GMT
References: <542@acf3.NYU.EDU>
Reply-To: roger@hotlips.UUCP (Roger Scott)
Distribution: na
Organization: Data General Corp., Sunnyvale, CA.
Lines: 43

In article <542@acf3.NYU.EDU> pedersen@acf3.UUCP (paul pedersen) writes:
>Sorry to post this, but megatest.UUCP and riacs.ARPA are both unknown to my
>mailer.
>
>In article <275@goofy.megatest.UUCP> you write:
>>
>>A while back, I asked what bad things would happen if I hacked cfront ...
>
>The reason for prefixing field names with struct names is to handle old
>C compilers, ...
>
>I can't account for the "au0_" prefixes, but probably the reason is something
>similar: handling old C compilers.
>

Sorry, but this explanation is wrong on both counts.  Structure member names
are prefixed with the structure name in order to handle the following case:
	struct base {
	    int bb;
	};
	struct derived : base {
	    int bb; // perfectly legal
	    int dd;
	    void f();
	};
	void derived::f() {base::bb = bb;}

base::bb and derived::bb can't very well be called the same thing in the C
code.  If you do not use this marginally useful ``feature'', you can make
a trivial change to print.c to fix this.  If you don't have source access,
but you are adventuresome, you can modify the symbol table info in the
a.out file (for dbx) [write me if you want to know more].

As for _au0, etc., it is used to handle a similarly marginal feature:
	int x;
	void foo() {
	    int x = ::x; // remember the ol' scope resolution operator?
	    // ...
	}

Again, if you don't use this you can eliminate the _au0.

p.s. - if you think things are bad now, wait 'till you see what 2.0 does
to names.