Preaching to the already converted: Om

So there are lots of things people here have already been exposed to, but a lot of them seem to fail to get much traction. Thus I very much appreciate this talk by David Nolan which helps "sell" more of the FP side of the FP vs. OO wars if you will, via Om: Clojurescript for React.

(Of course, he has to go and spoil it all by (1) selling AOP as well, and (2) insufficiently stressing the frustrations of trying to do these things in the Real World... but I guess I can't expect perfect nirvana to descend upon all people everywhere.)

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

AOP + React is powerful

AOP + FRP is actually quite nice: AOP exposes a lot of input streams that otherwise would require pretty global rewrites or heavy lifting like monadic interpreters. My favorite academic example of this is MzTake for declarative debugging/monitoring, and in the real world, tools like DTrace.

up to a limit

the problem i have is that it is not a good general solution. and yet people get all in love with it and then we have to live with the results? there are probably more principled ways of going about getting the same leverage. so yeah, i love me some dtrace, fine, but that doesn't mean i think e.g. aop for security concerns or whatever really ends up with a code base i would ever enjoy having to grok and maintain.

He's advocating aspects for narrow purposes

Namely: debugging and introspection.

If you look at Nolen's work, you'll see all his uses of Om's instrument are for little inspectors, profilers, etc.


the way the talk was presented, it was never stated that there's any rational, measured limits to using AOP. the absence of such statements is quite criminally misleading at the very least, i feel; otherwise the person DOESN'T EVEN KNOW that AOP has problems, which is even worse.