It's all spinning wheels and self-doubt until the first pot of coffee.

Thoughts on Pipes on the Web - Part II

In the last post, I expressed concern for my own well-being over a lack of head-over-heels love for Yahoo! Pipes. On the surface, I think it's because it's not a freshly discovered concept for me, and also probably because I'm tired and a bit hacked out right now. But, I think there's a bit more to it.

First of all, I'm not impressed by GUI builders for most things. Don't get me wrong: That Pipes GUI is pretty sweet and quite an impressive implementation — but as a rule, such things don't quite get my geek on. So, there's that.

Another thing is that I've done something kinda like Yahoo! Pipes, called FeedMagick — only, it doesn't have a GUI and I mostly abandoned it after releasing the code and having used it for a project at my old job. It's one of those serial enthusiasms that I've figured I'd circle back to eventually. I still think it's a pretty cool idea.

All kudos to the Yahoo! Pipes team though: Unlike me, they've actually got a living and breathing project — which trumps a paged-to-disk serial enthusiasm most days.

So what else is really curbing my enthusiasm? Well, Yahoo! Pipes looks like a pretty self-contained pipes engine — data goes in one end, frobnication happens in the middle, data comes out the other side.

But, what I like about the notion of the URL-line is that you can take one URL and supply it as a parameter to another URL — making messy pipelined URLs while building a crazy web-wide distributed execution environment powered by HTTP and REST. This is the kernel of the notion that I think really excites me about pipe on the web — I just haven't had a chance to do much with it lately.

Depending on the perspective, true pipes on the web — that is, URLs fetching URLs — look to me like functional programming ala Lisp. Consider an idempotent GET request as a pure function call with no side-effects. Then, consider a GET request that accepts a URL as a parameter — it's a nested function call: the outer GET must make an inner GET to fetch the parameter-supplied URL. Give the inner URL another URL as a parameter, and you've got yet another nested function call.

But, if you like, ignore the theoretical benefits of functional programming — flip the nested function calls inside out and you've got a pipe. And, since you're using HTTP GET, you can get all the benefits of HTTP — like caching of execution results and a web full of distributed processing nodes, among other things.

I haven't explored Yahoo! Pipes deeply enough yet, so maybe I'm missing the features that pipe authors can used to call on other distributed pipe elements out on the web at large. But, I think that's what ultimately gets me psyched about pipes the web and hasn't yet for Yahoo! Pipes.

And, of course, I think I'm just a little hacked out and tired of whizbang new stuff right now. :)

Archived Comments

  • The thing that struck me about Pipes is that it seemed rather limiting to someone who can code but too geeky to someone who can't. Maybe I am wrong but I wonder that once the hype dies down who exactly the audience will be long term.

  • Don't be so apologetic! Conceptually it's cool, but the UI quickly becomes frustrating for anyone with text-based coding experience. I guess this is the web's visual basic and we are all going to be more interested in it's perl.