Rachel Kroll
Wow, okay, the "feed reader score" post has been making some waves. There have been some good discussions about it, and very few of the usual failure modes for those forums.
I heard from a bunch of people who are feed reader authors or just feed reader users who care deeply about doing the right thing. They actually want to point their program at my proposed "score" site to see what it's doing, and so they can find out if it's behaving badly somehow.
I like hearing this! This kind of stuff warms my heart.
So, I've started taking steps to invest in this project. To that end, I ran out, had a couple of slices of pizza and some root beer, then came back and started digging in my parts box. I grabbed the old Raspberry Pi B+ (star of the armv6 post from last fall) and started bringing it up to date.
I got it all configured as a proper outside-facing box, then I actually *hopped in the car* earlier and drove it out to the data center to plug it in. It's there now, just waiting for me to start doing stuff with it.
Why a separate box? Well, I've done this kind of stuff before, and it's easy for a "test target" to turn into a smoking crater. This way, if it goes really badly, I can just turn it off and it won't affect the other boxes out there. It isn't affected by whatever rate-limiting I might do on the "real" web server.
Also, it caps the amount of resources that can be spent by this project - it's a goofy little box that'll barely move a few tens of Mbps across its godawful NIC. It's not going to be haul-ass fast or anything like that, but it doesn't have to be. It just has to log the incoming requests. Something else gets to do the analysis to figure out whether the observed behavior is good, bad, or just plain meh.
To be clear, the "feed" this thing will be serving is not going to have any real posts in it. You won't be able to read my latest stuff by subscribing to it. Also, I will warn everyone right now that I fully expect to have to yank the entire sub-zone out of the domain multiple times after people ruin one, and I have to NXDOMAIN them early.
The URLs you get for testing are not expected to be long-term durable, in other words. It all depends on how much abuse it gets. Considering you can figure out what a feed reader is doing after a day or two at most, that's probably not going to be a problem.
Why am I writing about this? I'm trying something different: basically, if I talk about it first, will that light a fire under me to make stuff happen sooner? This is not normally how I operate. I normally just show up with something already done. This kind of project is just big enough to where that won't work.
I guess it's time to ignite a petroleum product at zero-hundred hours.