Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!ucsd!sdcsvax!ucsdhub!hp-sdd!hplabs!hpda!hpwala!cfisun!ima!mirror!necntc!necis!mrst!sdti!turner
From: turner@sdti.SDTI.COM (Prescott K. Turner)
Newsgroups: comp.lang.c++
Subject: Re: C++ constructor and function semantics
Message-ID: <373@sdti.SDTI.COM>
Date: 20 Mar 89 22:26:07 GMT
References: <1149@oswego.Oswego.EDU>
Reply-To: turner@sdti.SDTI.COM (Prescott K. Turner, Jr.)
Organization: Software Development Technologies, Sudbury MA
Lines: 28

The notion that
      X var(fun());
should be equivalent to
      X var = fun();
is a red herring in this discussion -- there are other cases where cfront
makes a more blatant distinction between them (as when "fun" returns a type
which can be converted to X in 2 steps but not 1).

In article <1149@oswego.Oswego.EDU> dl@rocky.oswego.edu (Doug Lea) writes:
> (Along these lines, I wonder just how many existing C++ programs will
> `break' if/when cfront and g++ make `X var(fun())' operationally
> equivalent to `X var = fun()'.)
Such a change is not necessary, because the latter syntax will serve.
It is not desirable, because it complicates the meaning of 
      initializer:   ( expresson-list )
which means 
      Initialize the declared object using the constructor
      determined by the expression-list as a list of arguments.

In most circumstances, I think the elmination of calls to X(X&) is a good
thing.  To the extent that cfront and g++ handle it, it appears easier to
implement than other aspects of C++.  As Lea points out, users could
benefit a lot from standardizing on this behavior.
--
Prescott K. Turner, Jr.
Software Development Technologies, Inc.
375 Dutton Rd., Sudbury, MA 01776 USA        (508) 443-5779
UUCP: ...{harvard,mit-eddie}!sdti!turner    Internet: turner@sdti.sdti.com