From owner-sc22wg5@dkuug.dk Wed Jun 25 22:56:35 2003 Received: (from majordom@localhost) by dkuug.dk (8.12.8p1/8.9.2) id h5PKuZ2u061840 for sc22wg5-domo; Wed, 25 Jun 2003 22:56:35 +0200 (CEST) (envelope-from owner-sc22wg5@dkuug.dk) X-Authentication-Warning: ptah.dkuug.dk: majordom set sender to owner-sc22wg5@dkuug.dk using -f Received: from math.jpl.nasa.gov (math.jpl.nasa.gov [137.79.7.57]) by dkuug.dk (8.12.8p1/8.9.2) with ESMTP id h5PKuTEc061835 for ; Wed, 25 Jun 2003 22:56:31 +0200 (CEST) (envelope-from vsnyder@math.jpl.nasa.gov) Received: from math.jpl.nasa.gov (localhost.localdomain [127.0.0.1]) by math.jpl.nasa.gov (8.12.8/8.12.8) with ESMTP id h5PKuBJe002382; Wed, 25 Jun 2003 13:56:11 -0700 Received: from math.jpl.nasa.gov (vsnyder@localhost) by math.jpl.nasa.gov (8.12.8/8.12.8/Submit) with ESMTP id h5PKuB0r002378; Wed, 25 Jun 2003 13:56:11 -0700 Message-Id: <200306252056.h5PKuB0r002378@math.jpl.nasa.gov> X-Mailer: exmh version 2.4 06/23/2000 with nmh-1.0.4 X-Exmh-Isig-CompType: repl X-Exmh-Isig-Folder: inbox To: sc22wg5@dkuug.dk Cc: longb@cray.com Reply-to: Van.Snyder@jpl.nasa.gov Subject: Re: (SC22WG5.2824) Submodule association and host association In-Reply-To: Your message of "Wed, 25 Jun 2003 15:02:03 CDT." <200306251958.h5PJwZRQ061483@dkuug.dk> From: Van.Snyder@jpl.nasa.gov Mime-Version: 1.0 Content-Type: text/plain Date: Wed, 25 Jun 2003 13:56:11 -0700 Sender: owner-sc22wg5@dkuug.dk Precedence: bulk Bill Long wrote: > ... in a previous post the philosophy seemed to be "even if there is no > need for a feature we should not disallow it". I think the problem > with that philosophy is the difficulty in drawing the line between > "artificial restriction" and "feature bloat". > The other problem I see > is that the length of the feature list is roughly proportional to the > time you have to wait before an actual compiler implementation appears. I've only written little compilers for little languages -- never anything even remotely as ambitions as a Fortran compiler. So correct me if I'm wrong. It should be easier to allow a natural generalization of a feature by ignoring a gratuitous constraint than it is to enforce it. For example, many Fortran 77 processors found it easier simply to allow list-directed I/O for internal files than to go to the trouble to prevent it. The relation of a submodule to a parent submodule would be exactly the same as its relation to a parent module. So prohibiting a submodule's parent to be a submodule is an artificial restriction. The relation of a separate interface to its corresponding separate procedure body is the same no matter whether they're in the same program unit or the body is in a proper descendant of the one containing the interface. So prohibiting them to be in the same program unit is an artificial restriction. If there are sound reasons why implementation would be substantially easier with these restrictions, we need to hear them. Sure, the design of a processor is affected by the choice, but is it easier, or just different? For example, one may reason "Submodules are different from modules in that I cannot access a submodule by USE association. So I won't put submodule information into a .mod file; rather, I'll use a file with a different extension, and entirely different organization. If a submodule's parent can be a submodule, I'll need to use a .mod file in both cases, and put in a flag that says 'this is a submodule; USE association is not allowed." If I were a developer, I might do the latter anyway, just to save the effort of developing an entirely new separate-compilation infrastructure. Then a prohibition that a submodule cannot be the parent of a submodule would require extra work: I'd have to check that flag and generate a diagnostic both for the USE case and the parent case. I also suspect that having a .mod file for submodule information would lead to better diagnostics. The processor could say "The submodule FOOBAR cannot be accessed by USE association" instead of "Huh???" If the absence of a restriction doesn't cause a headache for developers, I always come down on the side that allows the greater variety of styles. I have never understood the urge to impose restrictions just so all programmers will be restricted to the subset of possible styles that one group finds to be sufficient. -- Van Snyder | What fraction of Americans believe Van.Snyder@jpl.nasa.gov | Wrestling is real and NASA is fake? Any alleged opinions are my own and have not been approved or disapproved by JPL, CalTech, NASA, Sean O'Keefe, George Bush, the Pope, or anybody else.