News You Can Bruise for 2002 December 17 (entry 2)

Tue Dec 17 2002 22:02 Call For Paper Suggestions: Some people at work (basically Josh) tried to get me interested in submitting a talk on NewsBruiser to the Python Community Conference, and sort of succeeded: I'm interested, but also apprehensive. Sometimes I consider submitting a talk to some conference or other, but I back out because I'm afraid of wasting people's time. Manoj pointed out that if people thought my talk would be a waste of time, they wouldn't show up. In that case, I responded, I was afraid of no one showing up. Ed pointed out that this would be the fault of the conference planners, for accepting a talk that no one wanted to attend. So I suppose I'm actually afraid of proposing an idea that will be rejected, which is a stupid reason for not at least giving it a try.[0]

So, I can think of a couple topics for a speech and discussion. I'd like feedback from you on what sounds interesting. I've included several ideas, along with the rebuttal provided by my internal critic. Background for those who need it: NewsBruiser is a web-based Python application of about 7500 lines which lets you run weblogs. Python is my favorite programming language; I use it a little bit at work and a lot at home, but I'm not an expert.

• NewsBruiser specifically - an overview. (This is lame and self-aggrandizing; you can just download it for an overview! Rebuttal to rebuttal: a lot of people like to learn about some piece of software by talking to the author about it. But it is self-aggrandizing.)

• Weblog software in Python, with NewsBruiser as a case study (This is a good idea, since I don't know of any other weblog software in Python except for PyBloxsom, so it's not something everyone knows about. The only problem is that this assumes that there's something inherently interesting about doing weblog software in Python, as opposed to doing some other kind of software in Python or doing weblog software in some other language.)

• It's easy to hide the complexity of those config files with a simple CGI interface! (Right now this is part of NewsBruiser; I'd need to separate it into a nice little includable library. But it's a good idea, and I could probably make it fun.)

• Writing self-contained Python applications and/or Python applications that run on old versions of Python. (Boooring. Besides, I don't have any systematic way of doing this; it's just a matter of choosing the right tools and writing to the right APIs.)

• Writing a program that's easy and fun to hack on. I've heard from several people (or seen it mentioned on their weblogs) that they downloaded NewsBruiser and hacked it to do such-and-so. This doesn't neccessarily speak well of NewsBruiser's not having accommodated them in the first place (though many old hacks have since been turned into core features), but it does speak well of NewsBruiser's hackability. What's my secret? I don't know, but I could try to find out in time for the deadline. (This seems like a good fit for the conference, so no real rebuttal here, but more than one case study would help with generalizations.)

• Picking the right abstraction: much well-designed software is great because it embodies a simplifying abstraction: witness UNIX's "everything is a file", or Plan 9's "everything is a file, and we mean it this time". Unfortunately, much poorly-designed software is awful precisely because it embodies the wrong abstraction. How to pick the right abstraction? (This could be fun, but I can't envision the people attending my talk having less experience than myself in picking abstractions.)

• Hiding UI complexity from those who didn't ask for it. NewsBruiser has a lot of features, and many features expose to the user knobs they can twiddle or forms they have to fill out. I choose defaults that expose as few knobs and forms as possible. (That's pretty weak tea, but it's all I can think of to say on the topic; it might be good as part of a larger talk, possibly the config interface talk.)

[0] Except, I might have an idea which the conference planners like but no one else does, so that no one shows up; or I might have an idea which sounds good enough to get people to show up but which actually is a waste of time. Also, I might show up to the talk in only my underwear, then suddenly be falling off of a cliff.