home *** CD-ROM | disk | FTP | other *** search
- Subject: Design Patterns/ODF?
- Sent: 6/27/96 5:26 PM
- Received: 6/27/96 5:41 PM
- From: William R. Jones, wjones@mindspring.com
- Reply-To: ODF Interest, ODF-Interest@CILabs.ORG
- To: OpenDoc Development Framework Discussion List, ODF-Interest@CILabs.
-
- Forgive me for resurrecting an ancient thread, but I only recently caught
- this while trolling through my mail archives.
-
- This Addison-Wesley book, "Design Patterns", is it about the OOD movement
- that has its roots in an architectural theory by Alexander? (Archibald?
- Atkinson? Something that starts with an A).
-
- Patterns, if I may simplify a lot, hold that just about every programming
- problem can be reduced to some easily-recognizable archetypes (i.e. stack,
- queue, on up in complexity to MVC, etc.), which then suggest or even dictate
- how the OO design should be constructed.
-
- It's a very powerful and elegant idea, rife with nuggets of truth. Bt what
- concerns me, and others, is that every single building project that used the
- original architectural technique was a dismal failure, patchworks that
- didn't combine into a useful whole.
-
- It can be argued, of course, that the field of architecture differs so much
- from computing that what didn't work there will work fine here.
-
- It may also be argued, though, that OOD/OOP lends itself to such patchworks,
- and further that component systems like OpenDoc and SOM do so as well. I
- think one of the biggest obstacles in selling component-based software is
- going to be convincing the market that it is a better approach than the
- allegedly organic design of monoliths like Word and Notes.
-
- So, if I'm building an application on OpenDoc components, on top of a
- patterns-like framework like ODF, writing in C++, which resolves to SOM
- binaries, I have, in effect, four layers that add tendencies toward
- patchiness, which makes me squeamish that I can make my application organic.
-
- In addition, I fret over a maxim that has yet to be proved false, in my
- experience-- that mapping an abstraction onto an abstraction onto an
- abstraction, etc., contributes to code bloat makes for slower systems. Which
- is why I know of no real-world implementation of the OSI 7-layer model.
- Exactly the opposite of the hopes I had for components in the first place.
-
- This just a concern I've had for a bit, but it bothers me enough to have
- considered not using ODF or C++ in my own code.
-
- Maybe I just spend too much time in engineering seminars.
-
- Bill
-