 Why OO Sucks
started 11/12/2002; 11:38:57 AM - last post 11/15/2002; 1:39:20 PM
|
|
Isaac Gouy - Why OO Sucks 
11/12/2002; 11:38:57 AM (reads: 828, responses: 14)
|
|
|
The thing I like about these is that they are about programming languages! It's taken for granted that "the world is OO" comments just show ignorance of ontology.
Why OO Sucks
- Data structure and functions should not be bound together
- Everything has to be an object
- In an OOPL data type definitions are spread out all over the place
- Objects have private state
and I liked Jonathan Rees' deconstruction of OO (although the hero-hacker pack-programmer thang always seems like smoke to me)
I'm fascinated that one concept is missing from these and most other pro/con accounts of OO: identity. Individuals (not objects) are important phenomena in any description of the world.
|
|
Jay Han - Re: Why OO Sucks
11/12/2002; 1:52:38 PM (reads: 881, responses: 0)
|
|
|
Both Armstrong's and Reese's critiques seem to be more about some language definitions and implications from those definitions. While interesting as they are about OOPLs and their "ill-effects", they are not really about OO (or rather OOAD).
BTW Henry Baker's paper on object identity is interesting:
http://home.pipeline.com/~hbaker1/ObjectIdentity.html
|
|
Isaac Gouy - Re: Why OO Sucks
11/12/2002; 2:23:46 PM (reads: 812, responses: 0)
|
|
|
While interesting as they are about OOPLs and their "ill-effects", they are not really about OO (or rather OOAD).
I agree.
Are you just pointing out that saying OO is a little sloppy - is that analysis, or design, or programming, or all of them - or do you mean something else?
|
|
Jay Han - Re: Why OO Sucks
11/12/2002; 3:43:50 PM (reads: 797, responses: 0)
|
|
|
I think Armstrong and Reese are somewhat correct about OOPLs. Their critiques are valuable. But they don't quite address harder issue of identifying "what, who and why" of a given problem domain. ("How, where, when" can be deferred a bit but never too long.) Obviously, no single approach has monopoly on addressing this general issue. But OOAD (and OOPL) over last 20 years have made a great deal of contribution on heuristics towards workable and usable solutions for this general issue. To me this legacy and still continuing works are valuable despite charges against OO for being sloppy or "mathematically unsound".
|
|
Isaac Gouy - Re: Why OO Sucks
11/13/2002; 12:05:43 PM (reads: 752, responses: 7)
|
|
|
So let's focus on OOAD, for a moment. Could you write the top 5 things that OO brings to problem analysis and to solution design?
(Seems to me we have to be specific just to make sure we are talking about the same things).
|
|
Ehud Lamm - Re: Why OO Sucks
11/13/2002; 1:14:29 PM (reads: 785, responses: 6)
|
|
|
LtU is dedicated to programming languages. If you want to go into real detail about design and analysis, please do so offline. The software used to run this site doesn't scale well, and is not really appropriate for long threads, anyway. Of course, discussing how OOAD is related to OOPL is an apppropriate topic for LtU.
|
|
Isaac Gouy - Re: Why OO Sucks
11/13/2002; 1:43:55 PM (reads: 835, responses: 0)
|
|
|
is not really appropriate for long threads
Oh. That takes the discussion out of Discussions.
discussing how OOAD is related to OOPL is an appropriate topic
Both proponents and opponents happily conflate them to the confusion of all.
|
|
Jay Han - Re: Why OO Sucks
11/13/2002; 3:49:13 PM (reads: 805, responses: 1)
|
|
|
Ehud: apologies for going off tangent. [At least the Baker paper was about languages.]
Isaac: if you want we can take this offline either by mail (see the URL of my name for address) or in my blog.
|
|
Ehud Lamm - Re: Why OO Sucks
11/14/2002; 1:23:42 AM (reads: 818, responses: 0)
|
|
|
No problem. I agree with Isaac that the distinctions are often blurred. If we can make the clearer, we should.
|
|
Noel Welsh - Re: Why OO Sucks
11/14/2002; 5:38:51 AM (reads: 787, responses: 2)
|
|
|
The software used to run this site doesn't scale well, and is not really appropriate for long threads, anyway.
This is an interesting comment. I sometimes think LtU would be better as a wiki as it would enable us to link up all the discussions. For example, there are multiple OO critiques in the discussion list at the moment and it would be good to tie them together and also reference the previous discussions we've had on this topic. I also believe that the discussion is what its really about. Anyone can grab papers off CiteSeer. It's the contextualising that adds value. All in my humble opinion. I don't intend to criticise LtU, just suggest improvements.
|
|
Isaac Gouy - Re: Why OO Sucks
11/14/2002; 9:17:02 AM (reads: 717, responses: 0)
|
|
|
Why OO Sucks seems to have the wrong title.
Why Abstract Data Types (in all their manifestations) Suck seems a better description.
|
|
Ehud Lamm - Re: Why OO Sucks
11/15/2002; 3:00:54 AM (reads: 810, responses: 0)
|
|
|
Wikis and weblogs each have their strong points. The good thing about the way this weblog is run (IMHO) is that papers and sites recommended by editors are almost lways very good, and worth checking out. This makes life easier, esp. if you don't have too much spare time. The weblog gives us levels of trust, which are useful for deciding what to read.
Wikis are great as repositories of intr-linked information.
It would great to have a wiki attached to LtU (sort of like what decafbad does). Alas, I don't have the facilities to run something like that.
LtU is based on the free software and hosting provided by Userland (thanks!) I find the system very comfortable, but it is not based on my design, so I am not always able to explain why things were designed the way they are. But, overall, LtU is ruuning for a couple of years now, and I find the setup very easy and applicable for the uses I had in mind.
More later.
|
|
Isaac Gouy - Re: Why OO Sucks
11/15/2002; 12:56:16 PM (reads: 681, responses: 1)
|
|
|
The software used to run this site doesn't scale well, and is not really appropriate for long threads, anyway.
There are discussion threads I'd like to respond to, but I'm concerned about feeding long-discussions that may cause problems on the LtU site.
Do we have any heuristics, or any work arounds? Like - spawn a new discussion after 15 postings?
Incidentally, I've found this to be an extremely tangible reminder of the point made about the value of massive concurrency in Erlang (How can it be concurrent if you're limited to 100 threads?).
|
|
Ehud Lamm - Re: Why OO Sucks
11/15/2002; 1:39:20 PM (reads: 706, responses: 0)
|
|
|
It's not really a problem with long threads. The real problem, I think, is when the structures of the thread becomes too deep (reply to a reply, and then out again etc.). Wehn you look at the whole thread, you don't see the structure. When you look at an individual message, you see the replies to it, but again you lose some context.
For almost all the discussions we had here there was no real problem. But when people start debating small issues, following the discussion becomes annoying. So each issue is often better served by starting a new thread.
|
|
|
|