From LPMeissner@msn.com Fri Mar 7 18:57:16 1997 Received: from upsmot03.msn.com (upsmot03.msn.com [204.95.110.85]) by dkuug.dk (8.6.12/8.6.12) with ESMTP id SAA04939 for ; Fri, 7 Mar 1997 18:57:13 +0100 Received: from upmajb06 ([204.95.110.89]) by upsmot03.msn.com (8.6.8.1/Configuration 4) with SMTP id JAA02086; Fri, 7 Mar 1997 09:50:52 -0800 Date: Fri, 7 Mar 97 17:53:42 UT From: "Loren Meissner" Message-Id: To: sc22wg5@dkuug.dk Cc: hebert@soleil.serma.cea.fr Subject: FW: Objects in Fortran 2000 -- AFNOR proposal Thank you for calling attention to the AFNOR paper. I am forwarding your message to the WG5 mailing list (which also includes all X3J3 members). = [The following observations are my own as an individual, and do not necessarily reflect those of any WG5 member or of any other X3J3 member. It was primarily in my role as Fortran Forum editor that I posted some information concerning Objects to the newsgroups. Note that in this role I never hesitate to take editorial positions, as well as to report factual observations, concerning the actions of Standards bodies and of others who have influence on the future of Fortran.] = It is my understanding that no definite action on Objects was taken at the Dresden meeting, because it was felt at that time that none of the proposals in this area had been the subject of sufficient prior consideration. As a result, I assume that the AFNOR paper became a part of the background, along with other papers in the WG5 and X3J3 repositories, for recommendations made by the "ad hoc electronic subgroup on DATA" that was established at Dresden and charged with reporting prior to the February joint WG5/X3J3 meeting in Las Vegas. = The Objects requirement established at the February meeting proposes a definite, although loosely defined, model. Some of the underlying principles include: Keep Objects in Fortran 2000 to a minimal level; Provide extension of derived data types ("single inheritance"); Provide run-time choice among "methods" based on the specific run-time type of an object (within its inheritance hierarchy). Considerable work remains, to refine this requirement and to establish the specific technical requirement. Under the current "division of labor" between WG5 and X3J3, this refinement is to be done by X3J3, with review by WG5 only to ensure that it is consistent with the established requirement. (I assume that WG5 could make further suggestions as to the direction of technical work in this area, but "micromanagement" would appear to be incompatible with the current separation of roles between the two groups.) = I have not closely followed any of the detailed deliberations in this area, either at Dresden or in February at Las Vegas, and I am not intimately familiar with the AFNOR proposal nor with the ADA-95 Objects facility, but it is my impression that the direction implied by the February action is consitent with (or at least, "not inconsistent with") the ADA-95 approach. = At recent meetings, both WG5 and X3J3 have expressed strong support for including Objects in Fortran 2000. Until recently, very little progress had been made in this area due to lack of agreement on the form that the facility should take. Some persons who have joined X3J3 since the beginning of 1996 have made considerable contributions to this discussion and to the current focus on a particular approach. It now seems possible to complete the development of Objects within the current F2000 schedule. = The next stages of the work are assigned to the X3J3 arena, and any specific contributions that you (or others) could make to this work can most profitably be directed to X3J3. As I have been pointing out for at least 20 years, little influence on any specific proposal is possible without diligent and persistent interaction with - i.e., close participation in - the group of individuals who spend large amounts of time at X3J3 meetings (and, to some extent, between meetings) considering detailed alternatives. Very little can be accomplished without consistently attending the meetings and interacting in good faith with the others who are working toward the same goals. = Personally, I fully agree that Fortran 2000 must reflect the progress that has been made in the computer world over the past couple of decades, and that this includes recoginition of Objects as an important aid to program reliability and maintainability. While Objects can be simulated to a large extent with the features of F90 and F95, I agreem that a few specific facilities should be added to F2000 to simplify and clarify the use of Objects. I am glad to see that some individuals and some proposals have come forward, so that it now seems that this will happen. = Loren P. Meissner -----Original Message----- From: hebert@ajax.serma.cea.fr Sent: Friday, March 07, 1997 5:30 PM To: Loren Meissner Cc: comp-fortran-90@mailbase.ac.uk Subject: Re: Objects in Fortran 2000 -- AFNOR proposal I would like to remember that AFNOR (the national standardisation body in France) made an official proposal to WG5 about Object Orientation in Fortran-2000. This proposal can be recovered from: ftp://ftp.ncsa.uiuc.edu/sc22wg5/ftp.nag.co.uk/N1151-N1200/PostScript /N1188.ps.gz The recommendation is largely based on the ADA-95 approach. Thank you for considering this proposal Alain Hebert hebert@soleil.serma.cea.fr