My Own Little World

Based upon Jon Udell’s writing and some of the work that has been happening here on Service-Oriented Architecture, I have come to the conclusion, which I am sure is not original but new to me, that this new greasemonkey/ajax/opensearch kabal is providing a poor-man’s implemenation of SOA. So no you don’t get all the fancy stuff that SOA gives you, but what you do get is cheap and relatively easy to implement. It’s not robust; if a web page designer changes the text or layout on the page you are currently ripping apart your greasemonkey script may no longer work. But then again, you are probably not using this to guarantee transactions with some client or relying on this for an industry workflow.

The real boon here is that the technology you have to use is easy to comprehend and there is no charge for access to the services. Most web coders understand html and the DOM, they understand xml and so the basic ideas are easy. The coding is not always easy because some pages are just plain broken, but your problem becomes a clever regex problem rather than having to learn a whole new technology. Since you are controlling the output from the services it becomes quite easy to string together.

I wrote a A9 opensearch hack to plug-in our EDN search page(go here and search for "garden of eden", read here to see what it’s about). The hardest part of the task was parsing the html, which isn’t that hard. When you are finished you have a url thats produces an RSS feed for a search. You can put that feed in an RSS reader, you can consume it with another application, you can put it in your pipe and smoke it. As soon as I was finished, I started thinking about stringing these types of searches together. For example you could take the output from my edn search, the output from a google search, and interweave them so you got EDN #1, G #1, EDN #2, G #2…
The problem has been transformed from how do I get my hands on the information to how do I best show the information I now have at my disposal. You have lego for information, this rocks.

One of my other side projects lately has been tech. reviewing Dave Johnson’s upcoming book on blogs, wikis, and feeds in action. And I am loving this book. While the code examples are nice I am more excited by the different ways Dave helps you think about using RSS and feeds to make your life better. And as a programmer he is also basically giving me more lego blocks to work with. My application produce feeds, which in turn can be integrated with other peoples feeds, and consumed as other programmers see fit. Given that the new RSS "specs" allow for binary objects (for things like podcasting), you have almost 80% of the uses anbody could want.

I have also been working on refactoring a WMS example at work. And then I remember how cool I thought the whole idea of WMS (warning this link opens a PDF) was from the beginning. I write my application so it can take a standard request and I will give you back an agreed upon image matching your request. No fancy web services, no CORBA, RMI, DCOM…just http. I am sure there are problems with this approach but it is still very appealing to me.

At least one problem I see with this is that you could potentially use peoples pages without ever laying an eyeball on their page. Since a lot of sites still count on eyeballs (and especially those eyes that produce click-throughs) for revenue this still of reorganization could cause problems. This is the same problem brick and mortar versus online. While amazon’s reviews of books are great, it still doesn’t replace crackin open the book and seeing if it’s really what I want. I think the irony is that this style of repurposing other people’s page is another notch up the foodchain from online stores. I also think its interesting that amazon is not outwardly afraid of this, since they even give you a toolkit to mine their DB.

If I were more marketing savy I would start figuring out how to take adavantage of the possibilities of this recombination. BTW some more props need to go out to various other people’s blogs but I can’t remember which ones were relevant to this posting.