OO-DBMS
-
So your argument is that OO is bad because you do it badly? I would take the exact opposite position on just about every point. OODBMSs are far more flexible and RDBMSs. (Usually, you're not required to worry about or maintain the schema.) One of the strengths of OO is its ability to model and use real-world data and behavior. OO is generally more applicable to business objects than UI controls. And inheritance, if done properly, is a wonderful abstraction. In many cases where there is conditional logic--and in most cases where there is a type cast--inheritance would have clarified and simplified the code.
JLengi wrote:
So your argument is that OO is bad because you do it badly?
No, my argument is that OO is bad because it doesn't correctly capture the real world.
JLengi wrote:
One of the strengths of OO is its ability to model and use real-world data and behavior.
Ah, that's where I disagree. The real world is a constantly changing environment of data, behavior, and relationships. One of the first things I learned about OO is that it is wonderfully appropriate when modeling extremely well defined and static relationships, and very poor, except in terms of abstractions, of modeling dynamic relationships. I can point to several records management systems I've worked on over the years that became calcified in their inability to adjust to new relationships in the data that the customer wanted to see, because OO was used originally to hardwire the customer's requirements 15 years ago, which of course have changed, and the database also reflected those hardwired initial conditions. If OO hadn't been used for those, what were then "real-world" data and behaviors, but instead a framework supporting dynamic data and relationships had been used (with sparse OO to support that), then the application could have continued to survive changing customer requirements.
JLengi wrote:
OO is generally more applicable to business objects than UI controls.
Again, I disagree, this time strongly. Business objects are not stable over time.
JLengi wrote:
And inheritance, if done properly, is a wonderful abstraction.
Agreement there. Interfaces are wonderful, and as you point out, conditional logic and type casts can be eliminated with proper OO modeling - that is one of its strengths. Marc
Testers Wanted!
Latest Article: User Authentication on Ruby on Rails - the definitive how to
My Blog -
JLengi wrote:
So your argument is that OO is bad because you do it badly?
No, my argument is that OO is bad because it doesn't correctly capture the real world.
JLengi wrote:
One of the strengths of OO is its ability to model and use real-world data and behavior.
Ah, that's where I disagree. The real world is a constantly changing environment of data, behavior, and relationships. One of the first things I learned about OO is that it is wonderfully appropriate when modeling extremely well defined and static relationships, and very poor, except in terms of abstractions, of modeling dynamic relationships. I can point to several records management systems I've worked on over the years that became calcified in their inability to adjust to new relationships in the data that the customer wanted to see, because OO was used originally to hardwire the customer's requirements 15 years ago, which of course have changed, and the database also reflected those hardwired initial conditions. If OO hadn't been used for those, what were then "real-world" data and behaviors, but instead a framework supporting dynamic data and relationships had been used (with sparse OO to support that), then the application could have continued to survive changing customer requirements.
JLengi wrote:
OO is generally more applicable to business objects than UI controls.
Again, I disagree, this time strongly. Business objects are not stable over time.
JLengi wrote:
And inheritance, if done properly, is a wonderful abstraction.
Agreement there. Interfaces are wonderful, and as you point out, conditional logic and type casts can be eliminated with proper OO modeling - that is one of its strengths. Marc
Testers Wanted!
Latest Article: User Authentication on Ruby on Rails - the definitive how to
My BlogMarc Clifton wrote: The real world is a constantly changing environment of data, behavior, and relationships. One of the first things I learned about OO is that it is wonderfully appropriate when modeling extremely well defined and static relationships, and very poor, except in terms of abstractions, of modeling dynamic relationships. I agree about the dynamic nature of the world. I disagree that OO can't model it. I spent a total of 15 years of my career working at two companies on two COTS product suites, one a family of model-based systems engineering tools, the other a family of knowledge management SDKs. In both cases, we used OO in all its glory to implement an entity-relationship-attribute (including attributed relationships) meta-model. Entities, relationship, and attributes were instances. The types/definitions of those entities, relationships, and attributes, were also instances. Entity definitions could even inherit from other entity definitions. Elegant and maintainable. I have also seen designs that suffer from the maladies you describe.
-
Marc Clifton wrote: The real world is a constantly changing environment of data, behavior, and relationships. One of the first things I learned about OO is that it is wonderfully appropriate when modeling extremely well defined and static relationships, and very poor, except in terms of abstractions, of modeling dynamic relationships. I agree about the dynamic nature of the world. I disagree that OO can't model it. I spent a total of 15 years of my career working at two companies on two COTS product suites, one a family of model-based systems engineering tools, the other a family of knowledge management SDKs. In both cases, we used OO in all its glory to implement an entity-relationship-attribute (including attributed relationships) meta-model. Entities, relationship, and attributes were instances. The types/definitions of those entities, relationships, and attributes, were also instances. Entity definitions could even inherit from other entity definitions. Elegant and maintainable. I have also seen designs that suffer from the maladies you describe.
JLengi wrote:
In both cases, we used OO in all its glory to implement an entity-relationship-attribute (including attributed relationships) meta-model.
Exactly! When it comes to meta-modeling, I think OO is a great solution. I've done very similar E/R/A systems in the past. Marc
Testers Wanted!
Latest Article: User Authentication on Ruby on Rails - the definitive how to
My Blog -
As we love OO designs in other fields then why did Object-oriented DBMS are not popular?
1010111011
Monster Maker wrote:
As we love OO designs in other fields
Pretty sure that my hammer isn't OO. Neither is the keypad on my calculator. Nor for that matter is IP traffic. The idiom applies to a small set of real world solutions and databases are not one of those. For the vast majority of cases, the utility of objects in a database do not match the needs of processing data (not objects) in the database.
-
Probably because they don't play nicely with SQL. SQL is very often what management think they understand so it's an unarguable requirement. The ORM or Object Relational Mapping is the messy expensive compromise that lets you pretend you're dealing with an OO DBMS as a developer but allows SQL to contnue to be used.
"The secret of happiness is freedom, and the secret of freedom, courage." Thucydides (B.C. 460-400)
Matthew Faithfull wrote:
Probably because they don't play nicely with SQL. SQL is very often what management think they understand so it's an unarguable requirement
Wrong. Technologies that offer actual benefits are embraced rapidly across wide spectrums of development. Management seldom makes technology decisions unless that is their job (like a CTO.) Real business managers care about cost, delivery and sales. Technologies that do not provided benefit or which prove detrimental are rejected. And there are many more of those than successes. And it is easy to find examples of failures and successes. "Object Oriented Programming" itself is one of the success cases. CASE Tools as well as OO databases are ones that failed to gain any significance in the development realm because they do not provide benefits. And OO databases have had decades in which to gain acceptance and have not.
-
And here's one example of why we do need OOP in the real world. I'm currently refactoring a library of a few dozen classes. This job is taking approximately twice as long as it otherwise would because the original author although a undoubtedly a genius did not bother over much with that basic tennet of OOP encapsulation. The library is full of public data members and friend declarations so that whenever I change the internals of class I have to trawl the whole library fixing errors all over the place where code is diving into the internals of other classes and modifying unencapsulated data. Now just imagine how much fun I'm going to have when I want this code to be thread safe. +5 for OOP :)
"The secret of happiness is freedom, and the secret of freedom, courage." Thucydides (B.C. 460-400)
Matthew Faithfull wrote:
And here's one example of why we do need OOP in the real world.
I don't see anything in that description that suggests it has anything to do with OOP. Poor implementation is poor implementation. If anything your description suggests a failure at process control which is a people problem not a technology problem. Or at least on might presume that if stringent code reviews had been in place the quality might have been better.
Matthew Faithfull wrote:
Now just imagine how much fun I'm going to have when I want this code to be thread safe.
That of course has absolutely nothing to do with OOP nor perhaps even process control. If the original app didn't need threads then it would have been unprofessional to write the code to support threads. The fact that requirements change over time is part of business and nothing more than how the real world works.
-
I used db4o[^], an object database, on a project a few years ago. Experience was mixed. What query language do you use? There wasn't a standard, and this object database had 3 different query languages. How do you handle unbounded result sets? Object databases have to guess how deep to load objects. It's kind of messy. That said, I find SQL terrible for today's apps: data is stored exactly once, so that writes are fast and reads are slow (usually requiring multiple joins or multiple queries, or both). That worked for 1970s when disk space was at a premium. Now it makes sense to optimize for reads; store the data multiple times if you need so that reads are blazing fast. Let writes be slow, since our applications tend to read far more than write. For these reasons, I'm using RavenDB[^]. It's a non-SQL database built for .NET developers. First class LINQ support, simple API, no tables or schemas needed. I like it.
My Messianic Jewish blog: Kineti L'Tziyon My software blog: Debugger.Break() Judah Himango
Judah Himango wrote:
so that writes are fast and reads are slow (usually requiring multiple joins or multiple queries, or both).
So you are claiming that that is the idiom needed for NMS collection system which collects statistics from a vast array of elements for a large corporation? Say where the TPS could exceed 100,000? And where data is accessed solely via summarized static aggregations (with a trivial amount of drill down)?
-
Monster Maker wrote:
As we love OO designs in other fields then why did Object-oriented DBMS are not popular?
Because I don't love OO designs. The concepts are restrictive and all too often inappropriately applied. A relational DB is simple and flexible, one doesn't care if a foreign key represents a "has a" or "is a kind of" relationship -it's simply a relationship. Furthermore, OO all too often ends up representing more how the programmer wants to work with the data than the actual true relationships of the data--think denormaliation. So, mapping a denormalized or otherwise bad OO design onto a relational DB is Not A Good Idea. Finally, given that I found very little use for OO other than UI controls and some limited abstractions, I can't imagine why I would think an OO-DBMS would be a useful thing. Interfaces are useful, classes are great as containers of behavior and state, and that's about it. All those inheritance graphs? X| Marc
Testers Wanted!
Latest Article: User Authentication on Ruby on Rails - the definitive how to
My BlogMarc Clifton wrote:
Furthermore, OO all too often ends up representing more how the programmer wants to work with the data than the actual true relationships of the data
However to be fair that is often how data modeling ends up represented in the database as well. Especially with Agile.
Marc Clifton wrote:
All those inheritance graphs
Not sure to exactly what you are referring but if that refers to either OO database base, or data model or even OO programming and where "all" means a whole lot then there is something wrong with the design. None of the mentioned items should have a lot of inheritance. That suggests design problem where a developer thinks that an Employee and Sales Contact are both People rather than recognizing that there is only Person with different properties.
-
Marc Clifton wrote:
all too often inappropriately applied
That's no reason not to love something! I've seen some very inappropriate database designs in my time -- but that's no reason for not loving SQL!
Marc Clifton wrote:
one doesn't care if a foreign key represents a "has a" or "is a kind of" relationship
If it's an HAS A it's a one-to-many relationship. If it's IS A KIND OF it's a many-to-one. So one does care- just using different terminology, really.
Marc Clifton wrote:
OO all too often ends up representing more how the programmer wants to work with the data than the actual true relationships of the data--think denormaliation.
Again, isn't this down to poor design (assuming it's done badly)? There are plenty of good reasons to denormalise a database - but many more bad ones. Again, I've seen some appalling messes of database design, and of OO design - but it's the designer at fault not the tool (grin) I'm genuinely surprised at someone not finding use for OO. When I was first introduced to OO (many many years ago) I really couldn't see the point - but the more I was forced to use it, the more I liked it, to the point where I started thinking in OO design terms - and when that happens everything just clicks into place. I genuinely don't think I could write good, non OO code any more - I'd guess that you use classes and objects at a single level as containers for methods and properties and not much more? This isn't a criticism - I know you have a wealth of experience - I'd be interested in seeing how you develop non-OO software these days!
MVVM # - I did it My Way ___________________________________________ Man, you're a god. - walterhevedeich 26/05/2011 .\\axxx (That's an 'M')
-
Marc Clifton wrote:
Furthermore, OO all too often ends up representing more how the programmer wants to work with the data than the actual true relationships of the data
However to be fair that is often how data modeling ends up represented in the database as well. Especially with Agile.
Marc Clifton wrote:
All those inheritance graphs
Not sure to exactly what you are referring but if that refers to either OO database base, or data model or even OO programming and where "all" means a whole lot then there is something wrong with the design. None of the mentioned items should have a lot of inheritance. That suggests design problem where a developer thinks that an Employee and Sales Contact are both People rather than recognizing that there is only Person with different properties.
jschell wrote:
That suggests design problem where a developer thinks that an Employee and Sales Contact are both People rather than recognizing that there is only Person with different properties.
That's where some of the ambiguity lies for me, and I remember pondering these issues a lot when first learning OO - should Employee and Sales Contact be treated as "is a kind of Person" or should they "have a" role: employee, sales contact, etc., (and of course, possibly multiple roles depending on their relationship to other entities.) There's reason enough to answer "both", as it depends on the circumstances. Ultimately, in my ROP articles by exploring the idea that the relationship itself is a first class denizen of the object world, capable of having its own properties. Marc
Testers Wanted!
Latest Article: User Authentication on Ruby on Rails - the definitive how to
My Blog -
So your argument is that OO is bad because you do it badly? I would take the exact opposite position on just about every point. OODBMSs are far more flexible and RDBMSs. (Usually, you're not required to worry about or maintain the schema.) One of the strengths of OO is its ability to model and use real-world data and behavior. OO is generally more applicable to business objects than UI controls. And inheritance, if done properly, is a wonderful abstraction. In many cases where there is conditional logic--and in most cases where there is a type cast--inheritance would have clarified and simplified the code.
JLengi wrote:
OODBMSs are far more flexible and RDBMSs
Nonsense.
JLengi wrote:
One of the strengths of OO is its ability to model and use real-world data and behavior
Good point. Care to explain exactly what business behavior you are putting into your OO database?
JLengi wrote:
OO is generally more applicable to business objects than UI controls.
Which has little to do with persisted data. Just as, for example, it has nothing to do with sending it over IP.
JLengi wrote:
And inheritance, if done properly, is a wonderful abstraction
Yes. It is however a shame that it is often misused.
JLengi wrote:
In many cases where there is conditional logic--and in most cases where there is a type cast--inheritance would have clarified and simplified the code.
Which is something that I almost never see. And I certainly don't remember seeing much of it when I did procedural programming.
-
Marc Clifton wrote: The real world is a constantly changing environment of data, behavior, and relationships. One of the first things I learned about OO is that it is wonderfully appropriate when modeling extremely well defined and static relationships, and very poor, except in terms of abstractions, of modeling dynamic relationships. I agree about the dynamic nature of the world. I disagree that OO can't model it. I spent a total of 15 years of my career working at two companies on two COTS product suites, one a family of model-based systems engineering tools, the other a family of knowledge management SDKs. In both cases, we used OO in all its glory to implement an entity-relationship-attribute (including attributed relationships) meta-model. Entities, relationship, and attributes were instances. The types/definitions of those entities, relationships, and attributes, were also instances. Entity definitions could even inherit from other entity definitions. Elegant and maintainable. I have also seen designs that suffer from the maladies you describe.
JLengi wrote:
...and very poor, except in terms of abstractions, of modeling dynamic relationships.
I haven't seen that. Dynamic data is exactly what static code manipulates. If there are meta models involved then at least to me, it obvious that the static code then manipulates meta models. The one difference of course is that it is generally harder to derive the required functionality required for the meta model and also to implement that as well. That however is due to complexity and not implementation.
-
jschell wrote:
That suggests design problem where a developer thinks that an Employee and Sales Contact are both People rather than recognizing that there is only Person with different properties.
That's where some of the ambiguity lies for me, and I remember pondering these issues a lot when first learning OO - should Employee and Sales Contact be treated as "is a kind of Person" or should they "have a" role: employee, sales contact, etc., (and of course, possibly multiple roles depending on their relationship to other entities.) There's reason enough to answer "both", as it depends on the circumstances. Ultimately, in my ROP articles by exploring the idea that the relationship itself is a first class denizen of the object world, capable of having its own properties. Marc
Testers Wanted!
Latest Article: User Authentication on Ruby on Rails - the definitive how to
My Blog -
Matthew Faithfull wrote:
And here's one example of why we do need OOP in the real world.
I don't see anything in that description that suggests it has anything to do with OOP. Poor implementation is poor implementation. If anything your description suggests a failure at process control which is a people problem not a technology problem. Or at least on might presume that if stringent code reviews had been in place the quality might have been better.
Matthew Faithfull wrote:
Now just imagine how much fun I'm going to have when I want this code to be thread safe.
That of course has absolutely nothing to do with OOP nor perhaps even process control. If the original app didn't need threads then it would have been unprofessional to write the code to support threads. The fact that requirements change over time is part of business and nothing more than how the real world works.
Remind me not to apply for a job at your place. We'll both be happier that way.
"The secret of happiness is freedom, and the secret of freedom, courage." Thucydides (B.C. 460-400)
-
Matthew Faithfull wrote:
Probably because they don't play nicely with SQL. SQL is very often what management think they understand so it's an unarguable requirement
Wrong. Technologies that offer actual benefits are embraced rapidly across wide spectrums of development. Management seldom makes technology decisions unless that is their job (like a CTO.) Real business managers care about cost, delivery and sales. Technologies that do not provided benefit or which prove detrimental are rejected. And there are many more of those than successes. And it is easy to find examples of failures and successes. "Object Oriented Programming" itself is one of the success cases. CASE Tools as well as OO databases are ones that failed to gain any significance in the development realm because they do not provide benefits. And OO databases have had decades in which to gain acceptance and have not.
Wrong. I used to work with an alternative database product, Multi dimensional OLAP, No SQL and the biggest problem in selling it was that managers install Oracle or SQL Server for the same reason they used to install IBM. The problem was not that it didn't work, it destroyed Oracle by 2 orders of magnitude in many applications but buyers wouldn't look at it because of there exisiting commitment to and investment in SQL technology. Maybe the USA is different but in the UK market having 'actual benefits' means very little at all.
"The secret of happiness is freedom, and the secret of freedom, courage." Thucydides (B.C. 460-400)
-
Wrong. I used to work with an alternative database product, Multi dimensional OLAP, No SQL and the biggest problem in selling it was that managers install Oracle or SQL Server for the same reason they used to install IBM. The problem was not that it didn't work, it destroyed Oracle by 2 orders of magnitude in many applications but buyers wouldn't look at it because of there exisiting commitment to and investment in SQL technology. Maybe the USA is different but in the UK market having 'actual benefits' means very little at all.
"The secret of happiness is freedom, and the secret of freedom, courage." Thucydides (B.C. 460-400)
-
JLengi wrote:
...and very poor, except in terms of abstractions, of modeling dynamic relationships.
I haven't seen that. Dynamic data is exactly what static code manipulates. If there are meta models involved then at least to me, it obvious that the static code then manipulates meta models. The one difference of course is that it is generally harder to derive the required functionality required for the meta model and also to implement that as well. That however is due to complexity and not implementation.
-
JLengi wrote:
OODBMSs are far more flexible and RDBMSs
Nonsense.
JLengi wrote:
One of the strengths of OO is its ability to model and use real-world data and behavior
Good point. Care to explain exactly what business behavior you are putting into your OO database?
JLengi wrote:
OO is generally more applicable to business objects than UI controls.
Which has little to do with persisted data. Just as, for example, it has nothing to do with sending it over IP.
JLengi wrote:
And inheritance, if done properly, is a wonderful abstraction
Yes. It is however a shame that it is often misused.
JLengi wrote:
In many cases where there is conditional logic--and in most cases where there is a type cast--inheritance would have clarified and simplified the code.
Which is something that I almost never see. And I certainly don't remember seeing much of it when I did procedural programming.
I wonder if you are having the same conversation I am having. I didn't say anything about putting behavior in the database. But that's kind of the point of an OODBMS, isn't it? If you can transparently persist your business objects as they are, then you don't need to worry about the interface between your application tier and your data tier. And there are serious advantages to executing your behavior as close to your data as possible, chief among them being performance. Fewer transformations in the case of RDBMSs. And less chattiness between tiers in the case of both RDBMSs and OODBMSs. There will almost inevitably be a proliferation of paths to the data. The "closer to the metal" your business logic is, the fewer places it will need to be implemented, and the less likely your validation/integrity rules will get inadvertently bypassed.
-
Monster Maker wrote:
As we love OO designs in other fields
Pretty sure that my hammer isn't OO. Neither is the keypad on my calculator. Nor for that matter is IP traffic. The idiom applies to a small set of real world solutions and databases are not one of those. For the vast majority of cases, the utility of objects in a database do not match the needs of processing data (not objects) in the database.
Perhaps. The utility of OODBMSs may be an order of magnitude less than that of RDBMSs. But the utility of OODBMSs is at least an order of magnitude greater than their current acceptance. (I.e., there are many applications that could and should use them but don't.) And the utility of OODBMSs would be an order of magnitude greater still if there were a little more investment in them (e.g., a standardized query language). The DBMS landscape has not reached equilibrium and is currently shifting away from relational and ACID.
-
When they say managed languages are slow, they never consider that managed languages are also OO languages. Then they put the blame on Garbage Collectors and on automatic memory management. I am starting to get tired of this shit. OO "architectures" are what is making the application get slow, endless series of patterns just for the new "agile" people be happy. Poor garbage collectors, they are excellent technology that free us from having to deal with memory cycle, just like when you have manual clutch on your car and get an automatic car. But they are getting all the burden of entire OO world alone.