From JLS@liverpool.ac.uk Mon Nov 9 10:50:18 1992 Received: from vm.uni-c.dk by dkuug.dk with SMTP id AA24436 (5.65c8/IDA-1.4.4j for ); Mon, 9 Nov 1992 20:16:17 +0100 Message-Id: <199211091916.AA24436@dkuug.dk> Received: from vm.uni-c.dk by vm.uni-c.dk (IBM VM SMTP V2R2) with BSMTP id 8570; Mon, 09 Nov 92 20:17:44 DNT Received: from UKACRL.BITNET by vm.uni-c.dk (Mailer R2.07) with BSMTP id 9730; Mon, 09 Nov 92 20:17:43 DNT Received: from RL.IB by UKACRL.BITNET (Mailer R2.07) with BSMTP id 9400; Mon, 09 Nov 92 19:15:51 GMT Received: from RL.IB by UK.AC.RL.IB (Mailer R2.07) with BSMTP id 7927; Mon, 09 Nov 92 19:15:50 GMT Via: UK.AC.LIV.IBM; 9 NOV 92 18:18:55 GMT Received: from UK.AC.LIVERPOOL by MAILER(4.4.t); 9 Nov 1992 11:22:07 GMT Date: Mon, 09 Nov 92 10:50:18 GMT From: Lawrie Schonfelder Subject: S20_122 ballot To: SC22/WG5 members X-Charset: ASCII X-Char-Esc: 29 I am slowly working through this in preparation for the ballot on items. Major works related to wholesale updating of the computing service I am responsible for have not left much time for F90 of late. I have so far two major problem areas. Items 6-11 are all related and probably should be combined (composited is the political party committee jargon) into a single clarification. There are internal contradictions in the existing set. There are clearly problems in this area which covers the multiple specification of specific and generic interfaces. I think the fix for the generic case is non-controvertial. The intent was that if more than one interface exists with the same generic-spec the effect is as if they were all combined into one block. If this produces an ambiguous overload then the program is in error. The question of multiple specification of the same specific interface is more complex. The intent was that as for data-objects this was to be not allowed, but I question whether this was a necessary restriction or even a desirable one. Given that there appears to be some ambiguity in the current text I think this question needs to be looked at very carefully again. There are many circumstances where it would be highly desirable for a given specific interface to be repeated (Maureen's F-word is just one). The primary need if this is done is for the interfaces so specified to be identical. This may be too big to handle as a CCI item but it is something to be considered for F96. I am totally opposed to the suggested fix for 13. The sentance at 5.3, p54, l29-30 "If a mapping is not specified for a letter, the default is the mapping in the host scoping unit." was most definitely intended to include interface blocks, and the examples on the next page are meant to be as they are. This was included as part of a public comment responce and I was one of the commentors. It was considered a terrible surprise for user if he were to put IMPLICIT NONE in his program but not in a interface declared in that program and so to suddenly find he had the system defined default mapping back. This above sentance was intended to apply to the implicit mapping by default for all contained scoping units. The user in may cases will not have the source for an external procedure. The declaration of the interface will have to be written in the context of the host program on the basis of the documentation. It will be the implicit mapping environment of the host the user is writing that is known not that of the external. If S20 comes forward as it is now I will be voting NO on these issues. Lawrie