# Ruby and AOP **URL:** \ **Category:** Ruby\ **Created:** [March 31, 2006, 4:56am UTC](https://www.ruby-forum.com/t/ruby-and-aop/93333 "2006-03-31T04:56:45Z")\ **Posts on this page:** 7\ **Page:** 1
**Author:** ![sjames](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/c57346/32.png) [@sjames](https://www.ruby-forum.com/u/sjames)\ **Post date:** [March 31, 2006, 4:56am UTC](https://www.ruby-forum.com/t/ruby-and-aop/93333/1 "2006-03-31T04:56:45Z")
I’m reading “The Pragmattic Programmer” right now and I’m at the part about orthogonality. The authors mention AOP and give an example involving log writing wherein you can “weave in” some AOP (“Aspect Oriented Programming” for those not in the know) sexiness and log writing occurs (the converse behaves as you might expect)\*\*. It seems to me that this behavior is _sort_ of like a mixin – i.e. you can mix it in for a new behavior. However, AOP code seems to be triggered when certain actions happen. Does this sound roughly correct to those who have AOP experience? Have any of you used AOP in your experience, and if so, where has it been most beneficial? Thank you, James H. \*\* For those interested, yes a Ruby AOP project exists: [http://aspectr.sourceforge.net/](http://aspectr.sourceforge.net/) ---
**Author:** ![sjames](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/c57346/32.png) [@sjames](https://www.ruby-forum.com/u/sjames)\ **Post date:** [March 31, 2006, 5:45am UTC](https://www.ruby-forum.com/t/ruby-and-aop/93333/2 "2006-03-31T05:45:23Z")
James H. wrote: > I’m reading “The Pragmattic Programmer” right now and I’m at the part > about orthogonality. The authors mention AOP and give an example > involving log writing wherein you can “weave in” some AOP (“Aspect > Oriented Programming” for those not in the know) sexiness and log > writing occurs (the converse behaves as you might expect)\*\*. It’s actually pretty much like this with aspectj (in java land). I haven’t used anything except vanilla aspectj, the new version is combined with AspectWorkz and uses @Annotation style java - something that so far I’ve not gotten into > It seems to me that this behavior is _sort_ of like a mixin – i.e. > you can mix it in for a new behavior. However, AOP code seems to be > triggered when certain actions happen. It’s not like a mixin in implementation > Does this sound roughly correct to those who have AOP experience? > Have any of you used AOP in your experience, and if so, where has it > been most beneficial? The proto-typical use case for AOP is logging, indeed the AspectJ team did a presentation where they demonstrated how much cleaner Apache Tomcat could be with an AOP approach towards logging - it’s quite an eye-opener how much code can be replaced by a single aspect. It’s all good in theory, but in practice (again in java land), the tools are slightly different ajc instead of javac, decding on runtime or loadtime weaving etc etc. Basically if you’re already bogged down with frameworks from hell (I’m looking at you JSF), then adding AspectJ/AOP in can be an extra thing to try and keep straight. That said, you can achieve some wonderful results. I added an auditing feature to a project where all methods that changed the state of data caused an aspect to fire and to record the persons name/change etc with a single aspect (\>100 lines of Java/AspectJ) - this made the code much cleaner as we didn’t have calls to an audit service before executing the business logic, we just let the aspect take care of that. I’ve also used it for security/authentication, again ensure the user is logged in before performing an action - another (relatively) straightforward aspect. One thing I’ve seen but never coded up is using aspects for caching (the example used JBoss cache and I didn’t have the willpower to download yet another app server just to test something) So erm, yeah aspects are quite cool, certainly worth looking at for a set of problems. IOC sort of stole their thunder and has negated the press AOP was getting for a while, but now most IOC frameworks use aspects internally. Kev ---
**Author:** ![sjames](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/c57346/32.png) [@sjames](https://www.ruby-forum.com/u/sjames)\ **Post date:** [March 31, 2006, 6:27am UTC](https://www.ruby-forum.com/t/ruby-and-aop/93333/3 "2006-03-31T06:27:53Z")
Another distinction is that AspectJ and related technologies use a “push” approach to weave aspects into different join-points in the code. In the case of a mixin, it’s more of a “pull” approach where you have to walk up to the type that you want to add the mixin functionality to. I spent a lot of time a few years ago implementing a runtime aspect-weaving extension to the CLR - that had a lot of potential - you could weave code into the middle of a running system, and then yank it back out when you were done. Great for doing things like injecting diagnostic probes into a corrupted system to perform post-mortem diagnosis. -John > **[John Lam's Blog](https://www.iunknown.com/)** > > Since 1997 Yes, it’s like a mixin but on more than a ‘class’ level. It’s ---
**Author:** ![sjames](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/c57346/32.png) [@sjames](https://www.ruby-forum.com/u/sjames)\ **Post date:** [March 31, 2006, 7:31am UTC](https://www.ruby-forum.com/t/ruby-and-aop/93333/4 "2006-03-31T07:31:48Z")
> So erm, yeah aspects are quite cool, certainly worth looking at for a > set of problems. IOC sort of stole their thunder and has negated the > press AOP was getting for a while, but now most IOC frameworks use > aspects internally. Pardon my naivity, but what is IOC? Thank you, James H. ---
**Author:** ![sjames](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/c57346/32.png) [@sjames](https://www.ruby-forum.com/u/sjames)\ **Post date:** [March 31, 2006, 6:03am UTC](https://www.ruby-forum.com/t/ruby-and-aop/93333/5 "2006-03-31T06:03:38Z")
On 3/30/06, James H. [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote: > Does this sound roughly correct to those who have AOP experience? Have > any of you used AOP in your experience, and if so, where has it been > most beneficial? Yes, it’s like a mixin but on more than a ‘class’ level. It’s cross-cutting(to use the jargon) across class hierarchies. Basically AOP allows a further and different separation of concerns in a software system. Think of transactions and security as better uses of AoP. An object should be able to participate in a transaction without ever knowing it’s been in one. Log writing is the ‘Hello World’ of AoP. I’ve used AoP for Logging and exception tracing at runtime of working systems. Ron Bodkin and Ramnivas Laddad have excellent blogs on AoP. hth, ---
**Author:** ![sjames](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/c57346/32.png) [@sjames](https://www.ruby-forum.com/u/sjames)\ **Post date:** [March 31, 2006, 8:29am UTC](https://www.ruby-forum.com/t/ruby-and-aop/93333/6 "2006-03-31T08:29:33Z")
On 3/31/06, James H. [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote: > > So erm, yeah aspects are quite cool, certainly worth looking at for a > > set of problems. IOC sort of stole their thunder and has negated the > > press AOP was getting for a while, but now most IOC frameworks use > > aspects internally. > > Pardon my naivity, but what is IOC? IOC = Inversion Of Control. See Marting Fowler’s article for a good explanation: [Inversion of Control Containers and the Dependency Injection pattern](http://www.martinfowler.com/articles/injection.html) There are several implementations in Ruby, Needle for example: [http://needle.rubyforge.org/](http://needle.rubyforge.org/) ---
**Author:** ![sjames](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/s/c57346/32.png) [@sjames](https://www.ruby-forum.com/u/sjames)\ **Post date:** [March 31, 2006, 8:29am UTC](https://www.ruby-forum.com/t/ruby-and-aop/93333/7 "2006-03-31T08:29:35Z")
Hi, James H. schrieb: > Pardon my naivity, but what is IOC? Inversion of Control. Read Martin F.'s article “Inversion of Control Containers and the Dependency Injection pattern” at > **[Inversion of Control Containers and the Dependency Injection pattern](https://www.martinfowler.com/articles/injection.html)** > > Explaining the Dependency Injection pattern, by contrasting it with Service Locator. The choice between them is less important than the principle of separating configuration from use. Lutz