From snir@watson.ibm.com Thu Jan 30 23:57:14 1997 Received: from igw3.watson.ibm.com (igw3.watson.ibm.com [129.34.139.18]) by dkuug.dk (8.6.12/8.6.12) with ESMTP id XAA17123 for ; Thu, 30 Jan 1997 23:57:12 +0100 Received: from mailhub1.watson.ibm.com (mailhub1.watson.ibm.com [9.2.249.31]) by igw3.watson.ibm.com (8.7.6/8.7.1) with ESMTP id RAA21920; Thu, 30 Jan 1997 17:57:50 -0500 Received: from watngi01.watson.ibm.com (watngi01.watson.ibm.com [9.2.235.20]) by mailhub1.watson.ibm.com (8.8.2/01-15-97) with SMTP id RAA29486; Thu, 30 Jan 1997 17:57:02 -0500 Received: by watngi01.watson.ibm.com(Lotus SMTP MTA v1.05b4 (287.3 12-16-1996)) id 8525642F.007E0EAB ; Thu, 30 Jan 1997 17:56:52 -0400 X-Lotus-FromDomain: IBM RESEARCH From: "Marc Snir" To: jcownie@dolphinics.com cc: Khb@Eng.Sun.Com, Malcolm@Nag.Co.Uk, Hennecke@Rz.Uni-Karlsruhe.De, Mpi-Bind@Mcs.Anl.Gov, Sc22Wg5@Dkuug.Dk Message-ID: <8525642F.007DA2D4.00@watngi01.watson.ibm.com> Date: Thu, 30 Jan 1997 17:56:43 -0400 Subject: Re: (SC22WG5.1282) PASS_BY("descriptor") Mime-Version: 1.0 Content-type: text/plain; charset=US-ASCII Contrary to what was claimed in a previous message, this type of functionality was standardized in HPF: HPF has a mechanism for passing array arguments with HPF distributions to non HPF code, without reshuffling or copying the arguments; the external (say, sequential Fortran) code has inquiry functions to find out how the HPF arrays are distributed. This does not require a standard representation of HPF dope vectors, but just inquiry functions that know how to interpret vendor dependent dope vectors.