Put religion in hands of people and they see it as a tool for power and control over others. Religious groups are afraid of science and scientific thinking eroding 'faith'. Faith after all is about believing in something that people do not understand and can't explain. More you learn, less is left for faith.
Intelligent design concentrates on things we do not fully understand or don't know about, and explains them with God. As curious creatures, people are seeking for answers and are ready to believe in something. Unaswered questions bother us to death.
Hmm, did I miss something. Why is it a big deal that some dinosaur ate both meat and veggies. Many animals do this today, including humans. Why would they turn to veggie-only diet then? Well, some people do too - vegetarians, heheh.
But jokes aside. Maybe veggies were just easier to get than meat? Maybe many of them ate meat when it was available, though rarely, and they just got 'lucky' finding one with meat.
Sure it is the point to choose the right tool, but according to some Ruby fanatics, it is always Ruby for everything. Ruby replacing Java, toasting J2EE, diminishing.NET - Kaboom! Illusions of grandieur. You have heard it. If not, go read some TSS threads where the fans are eroding any credibility that Ruby might have had.
There is always an excuse or explanation - it will be there in the next version, it is already beta, it is being planned, hype hype hype. The fact is that the enthusiastics would so very much like Ruby to be what Java is or what.NET is, but it is not - and they know it.
So they are creating incredible cover stories about invincible feats of low row counts and immense scalability, and then forget to turn the Java caching on when comparing applications, and btw, we can't see the application either just the compelling numbers - yes, Java empire has fallen!
It's good information, isn't it - better than plain hype and illusions of grandieur.
If Active Records solves different problems, why are people trying to compare them. And might I add that they are making a mess of it.
But I agree, it solves different problems and is not suitable to all tasks that Hibernate solves. As long as you remember that, you might be fine.
I am not saying RoR sucks, because it does not. I am just saying that people forget its limitations and try to apply it to situations where is does not work.
And finally, I thought we were talking about Ruby, not C. If Ruby has to be dependent on C because of performance reasons, it is trying to be more than it is worth.
Sure. Take an arbitrary database with weird field names that make no sense until after readin the documentation. Now let Ror map the fields to accessors and you get a nice alphabet soup that nobody can understand nor read. And guess what, most of the tables will be so abstract or do not have enough abstraction that the objects make no 'real-life' sense.
I am just saying that unless you as a developer are desiding what the database is going to look like, you are wasting your time with RoR.
RoR does not suck. People are just over-hyping it and jumping into conclusions.
RoR can not be successfully used with legacy databases that do not honor any kind of 'object' structure. I am not saying that Hibernate is any better, but you can not use RoR in this case, period.
As what comes to performance:
Active Record offers no viable caching. RoR has web tier caching but not data tier caching.
Active Record also loads everything lazily, which means that it suffers badly from N+1 loading problems. In other words it performs very badly with comples object graphs.
Also pure performance differences between Java and Ruby are clear. Take XML parsers for example, where Java is 50 times faster.
Having said that, you can use RoR for many things but don't think there are no severe limitations, because there are!
Active Record offers no viable caching. The test application might have had caching on the web tier, not the database tier.
Hibernate sweeps the tables with Active Record when it comes to caching and scalability. Hibernate supports two caching layers - which the author failed to use in his article as his numbers show, and what he openly admitted.
Active Record also loads everything lazily. Hibernate can do this but also offers other options. Lazy loading is one of the biggest fallacies of ORMs and will lead to n+1 loading problems. Since Active Record does not support outer joins, its performance is pitiful once the object graph grows. You have to write piggypack queries by hand with Rails to get the performance out and this gets extremely messy with larger databases.
Loading data lazily is good only if you did not need to load the data on the first round. This however is hardly ever the case when your object model is a little bit more complex. Very simple databases can have these problems as well depending on the object relations.
The test numbers obviously do not show pure Java vs. Ruby performance. The application is but a thin layer serving database stuff, and hence there is no significant 'logic'. But I can tell you that the fastest Ruby XML parser implementation is 50 times slower than Java.
That said, the author probably had a good reason to use RoR, which is fine. I am just saying that do not read too much into this.
Do not forget that he can not prove any of the numbers nor can anyone really dig in and verify any of his numbers, nor can anyone criticize his tests or get any idea what he was actually testing without the full source code available.
While the numbers probably mean something, they do not prove anything, nor should they be a reason for over-hyping this technology.
Don't read too much into this. Ruby is a nice language and Active Record is a nice ORM solution, BUT...
There are severe limitations for RoR. It can't work with legacy databases, since it relies on the object model based on the database tables and their relations. There is no caching of data either. Simplistic performance comparisons are very naive.
People are jumping into conclusions based on a tiny toy demo. It only proves that:
Ruby is good for small apps. It is better than Java in this scope, but only when you get to create the whole application from ground up.
There is no proof beyond that, and there won't be any time soon because Ruby does not have support for many critical features that larger applications require. Ruby is a scripting language, meant for scripting. Do not try to imagine anything else.
Slashdot Mirror
Intelligent design concentrates on things we do not fully understand or don't know about, and explains them with God. As curious creatures, people are seeking for answers and are ready to believe in something. Unaswered questions bother us to death.
First create unsecure software that lets the viruses in, then provide virus protection software, and make the user pay for both. Simply brilliant!
Hmm, did I miss something. Why is it a big deal that some dinosaur ate both meat and veggies. Many animals do this today, including humans. Why would they turn to veggie-only diet then? Well, some people do too - vegetarians, heheh. But jokes aside. Maybe veggies were just easier to get than meat? Maybe many of them ate meat when it was available, though rarely, and they just got 'lucky' finding one with meat.
There is always an excuse or explanation - it will be there in the next version, it is already beta, it is being planned, hype hype hype. The fact is that the enthusiastics would so very much like Ruby to be what Java is or what .NET is, but it is not - and they know it.
So they are creating incredible cover stories about invincible feats of low row counts and immense scalability, and then forget to turn the Java caching on when comparing applications, and btw, we can't see the application either just the compelling numbers - yes, Java empire has fallen!
If Active Records solves different problems, why are people trying to compare them. And might I add that they are making a mess of it.
But I agree, it solves different problems and is not suitable to all tasks that Hibernate solves. As long as you remember that, you might be fine.
I am not saying RoR sucks, because it does not. I am just saying that people forget its limitations and try to apply it to situations where is does not work.
And finally, I thought we were talking about Ruby, not C. If Ruby has to be dependent on C because of performance reasons, it is trying to be more than it is worth.
I am just saying that unless you as a developer are desiding what the database is going to look like, you are wasting your time with RoR.
RoR can not be successfully used with legacy databases that do not honor any kind of 'object' structure. I am not saying that Hibernate is any better, but you can not use RoR in this case, period.
As what comes to performance:
Active Record offers no viable caching. RoR has web tier caching but not data tier caching.
Active Record also loads everything lazily, which means that it suffers badly from N+1 loading problems. In other words it performs very badly with comples object graphs.
Also pure performance differences between Java and Ruby are clear. Take XML parsers for example, where Java is 50 times faster.
Having said that, you can use RoR for many things but don't think there are no severe limitations, because there are!
Hibernate sweeps the tables with Active Record when it comes to caching and scalability. Hibernate supports two caching layers - which the author failed to use in his article as his numbers show, and what he openly admitted.
Active Record also loads everything lazily. Hibernate can do this but also offers other options. Lazy loading is one of the biggest fallacies of ORMs and will lead to n+1 loading problems. Since Active Record does not support outer joins, its performance is pitiful once the object graph grows. You have to write piggypack queries by hand with Rails to get the performance out and this gets extremely messy with larger databases.
Loading data lazily is good only if you did not need to load the data on the first round. This however is hardly ever the case when your object model is a little bit more complex. Very simple databases can have these problems as well depending on the object relations.
The test numbers obviously do not show pure Java vs. Ruby performance. The application is but a thin layer serving database stuff, and hence there is no significant 'logic'. But I can tell you that the fastest Ruby XML parser implementation is 50 times slower than Java.
That said, the author probably had a good reason to use RoR, which is fine. I am just saying that do not read too much into this.
Do not forget that he can not prove any of the numbers nor can anyone really dig in and verify any of his numbers, nor can anyone criticize his tests or get any idea what he was actually testing without the full source code available.
While the numbers probably mean something, they do not prove anything, nor should they be a reason for over-hyping this technology.
Don't read too much into this. Ruby is a nice language and Active Record is a nice ORM solution, BUT... There are severe limitations for RoR. It can't work with legacy databases, since it relies on the object model based on the database tables and their relations. There is no caching of data either. Simplistic performance comparisons are very naive. People are jumping into conclusions based on a tiny toy demo. It only proves that: Ruby is good for small apps. It is better than Java in this scope, but only when you get to create the whole application from ground up. There is no proof beyond that, and there won't be any time soon because Ruby does not have support for many critical features that larger applications require. Ruby is a scripting language, meant for scripting. Do not try to imagine anything else.