Path: utzoo!utgpu!jarvis.csri.toronto.edu!ists!yunexus!geac!alias!rae%alias@csri.utoronto.ca
From: rae%alias@csri.utoronto.ca (Reid Ellis)
Newsgroups: gnu.g++
Subject: named return value [was: Defining operators]
Summary: ugly ugly
Message-ID: <650@alias.UUCP>
Date: 27 Nov 89 17:06:01 GMT
References: <1253@cnetlu.UUCP> <8911221001.AA29853@lurch.Stanford.EDU>
Sender: rae@alias.UUCP
Reply-To: rae%alias@csri.utoronto.ca (Reid Ellis)
Distribution: gnu
Organization: Alias Research -- the cutting edge in JAGGIE technology
Lines: 22

tiemann@lurch.stanford.edu writes:
|Why not allow a fourth form, using Named Return Values (see
|July/August C++ Report for details of this GNU C++ language
|extension):
|
|matrix operator+(matrix& x, matrix& y) return z { ... }

And look!  It's extensible!  Later we can add multiple return values!

matrix operator+(matrix& x, matrix& y) return x,y,z { ... }

Ya, dat's it..  :)

But seriously, what if operator+ is used inside an expression in this new
case?  Then the compiler generates a temp on its own, I guess..

e.g.:	A=(b+c)*2;

					Reid
---
Reid Ellis, 264 Broadway Avenue, Toronto ON, M4P 1V9, Canada
rae%alias@csri.utoronto.ca                   +1 416 487 1383


Brought to you by Super Global Mega Corp .com