I thought it was an ingenius first project for XML. The people I talked to about the idea before starting the project told me it was a horrible idea, and my professor agreed with them: I got 10 points taken off on the assignment because she called it a usability nightmare.

She just didn’t understand what I had done, and why it was so brilliant, if unfamiliar. She couldn’t see how if this was a “real” program, how useful this would be. In particular, if the user ran the program with a -v, -vv, -vvv, &c verbosity switch. Although I did not add support for logging to the program (diagnostic information was not required), I could have easily done so by adding a decorator that put diagnostic information about atomic data elements. Truth be told, though, at the time I wasn’t really aware about the consequences of my actions, and it was just the idea of “Hey! Let me learn XML in an Operating Systems course!” that happened to work out well.

Yeah, as you say there aren’t many easy-to-use frameworks that support XML Pipelines and that’s made it hard on developers. I think there’s been a problem over the years articulating their usefulness, which means that programmers haven’t thought of their web software as a component that consumes and publishes structured data.

Right now PilferPage is vapourware but I should get some time this week to put up some ideas for debate. So hang around and make sure that I don’t skew the language into a niche — it should be as generalised as possible.

I hadn’t read Sam Ruby’s post before, thanks for the link I’ll probably point that out next time I’m trying to explain this stuff.

At any rate, I agree that XML Pipelines are incredibly useful, yet underutilized in practice due to the amount of plumbing required to realize that usefulness. My first inspiration for XML Pipelines came when reading Sam Ruby’s blog posts about Planet Venus as a “river of news” feed reader. In particular, his blog post Genshi Filters for Venus (http://www.intertwingly.net/blog/2007/04/14/Genshi-Filters-for-Venus) was the first one where the value of an XML Pipeline woke me up. I didn’t understand what Sam Ruby was saying at first. It must’ve taken me 6 months to get it: It’s the model UNIX Pipes has been missing for complex event processing / stream event processing. Where UNIX Pipes stops at merely coroutines, the XML Pipeline picks up. Come to that, UNIX Pipes don’t provide the same level of data independence as the XML Pipeline. The benefit of robust data serialization is clear: The ability to identify and break off arbitrary nodes, re-purposing at will. Semi-structured is better than unstructured.

I’ve linked your blog (Entry #3) on the Andromeda Project mailing list, since the presentation layer of Andromeda is undergoing improvement and reconsideration. GUI API agnosticism for free would be ideal, as it would allow developers to take advantage of opportunity costs and the very fact it is agnostic could mean greater interoperability than Andromeda already possesses. Currently, Andromeda supports Joomla templates, and thus there are thousands of layouts freely available as well as an extra thousands more for a fee. Essentially, the goal should be to turn the front-end into a commodity, allowing more time for higher margin services. Being platform agnostic is a step in the right direction, but the end goal must be to make commodities out of the complements to those higher margin services. Hopefully, Andromeda developers can then simplify their cost estimates to customers.

Keep us informed.

You might want want to add planet.nztech.org to your “allow” list for linked images…

–Phil.