From jrbd@craycos.com Wed Feb 2 06:41:43 1994 Received: from ibmpcug.co.uk (Kate.ibmPCUG.CO.UK) by dkuug.dk with SMTP id AA01884 (5.65c8/IDA-1.4.4j for ); Wed, 2 Feb 1994 21:50:10 +0100 Received: from aspen.craycos.com by kate.ibmpcug.co.uk id aa05792; 2 Feb 94 20:42 GMT Received: from copper (copper.Craycos.COM) by aspen.Craycos.COM (4.1/TotalHack-4.1) id AA03827; Wed, 2 Feb 94 13:41:48 MST Received: from tnt by copper (4.1/SMI-4.1) id AA20146; Wed, 2 Feb 94 13:41:43 MST Date: Wed, 2 Feb 94 13:41:43 MST From: James Davies Message-Id: <9402022041.AA20146@copper> To: bill@amber.ssd.csd.harris.com, schmitt@edvz.tuwien.ac.at Subject: Re: (SC22WG5.493) Re: Comp.lang.fortran comments on evoluti Cc: sc22wg5@dkuug.dk X-Charset: ASCII X-Char-Esc: 29 >> >> > > The fact is, there is no technical reason for this restriction to >> > > remain. It is a holdover from ancient architectures. Furthermore, >> > > F90 now allows structures containing both CHARACTER and >> > > non-CHARACTER data, and COMMON is >> > > nothing more than a big structure. So what's the problem? >> >> > The difference is storrage association. >> >> No, it's not. We still have SEQUENCE structures that can be storage >> associated as long as the declarations are identical. No reason why COMMON >> blocks can't do the same. >> >But as you know, COMMON-blocks does not have the same declarations. I believe the entire problem with mixing CHARACTER with non-CHARACTER in common blocks or equivalences is defining how many characters there are in one integer or real, etc. 2, 4, 8, 10, ... ? I believe that the standard would have to define this in order for (especially) EQUIVALENCE to be portable, and there are obvious differences in current practice that couldn't be reconciled. Regards, Jim Davies jrbd@craycos.com