The Eighty-Twenty Solution

Went to the DALUG meeting last night. Andrew Dolbey, one of Larry Hunter's graduate students at UCHSC, gave a presentation on SOAP, in particular, the implementation provided by Franz. Andrew focused on the problems that still exist with interop among the platforms UCHSC uses, namely Lisp, Perl (SOAP::Lite) and Java. All these interop problems are using RPC/Encoded; Andrew pointed out that each language handles some types better than others. I was a little surprised to learn that Franz' implementation doesn't support doc/literal yet; my philosophy has been that doc/literal interoperates better than rpc/encoded ever will, but I guess rpc/encoded is what's in the spec and that's what people are writing to. Andrew's takeaway was that you can't assume things will interop, you have to test up front. This is a tractable problem for UCHSC, but seems like an ominous situation for people building generally consumable web services; however, it aligns with my own experience. I expect that Google and Amazon have similar experieces, given that they distribute client stubs for their services for the most popular languages. Andrew and Larry also discussed the various directories specific to bioinformatics, and touched specifically on the BioMOBY project. It turns out that there's a lot of work being done towards semantic web style services in bioinformatics. We had some discussion on this; when you get a room full of Lisp hackers together, somebody's going to remember the failed promises of AI and look askance on semantic web ideals. Larry made a great point; there's no common sense knowlege in Molecular Biology; everything you learned from a textbook, or from a paper you read, or from an experiment you did. Larry also mentioned that ontologies are a problem for biologists, but cited a few of examples of successful ontologies. However, the motivations for the participants are somewhat different as well; being researchers, biologists are very interested in disseminating their findings as widely and freely as possible; the whole idea is to have your research cited by others.

Last night, Bill asked me what was going on with the Scheme weblog code. Unfortunately, not much. I've been pretty swamped after vacation, and it's been hard to open up that code again. Right now, the main priorities are:

1. Optimize select-entries.
2. Respect <atom:issued> in the Atom posting API.
3. Implement the Atom API GET, PUT, and DELETE, especially PUT, so I can fix mistakes in my posts instead doing it on the server.
4. Implement trackback and/or pingback for new posts.

Also, looking at the Atom API, it looks like I have a few errors in how I implemented the create entry function. Anyway, there's a couple of good night's work there, and it's been difficult getting free time. Lately, I've wanted to do more CL, anyway. I've started reading On Lisp again as a way to get back into it. As for WordUp!, I'll get it in gear, one of these days.