v0.1.1 The romance of the telescope
Following a half-year or so break (I'll say half a year as my work between January and March was pretty minimal) I am probably able to work on Netfarm for a while again.
The main things I want to do are:
- make the Netfarm server more space efficient; it can pull up a lot of uninteresting objects in order to verify an interesting object. This is necessary to keep Netfarm "mostly trustless", but uninteresting objects can be held only temporarily in a LRU cache or something.
- use an ACID (emphasis on D for Durable) database; files apparently are hard and I am too stupid to keep them durable.
- recompute internal queues after stopping a server. I think they can be recomputed from the states we store already, but then again LRU caching adds transitions I didn't plan for... so I also have to
- check that the state machine we have going for object synchronisation and verification actually makes sense.
So, why the break? From memory I decided to "officially" declare that I needed a break a few days after publishing "Ethical software" is a sad joke. Typically it is easy to declare that some people just need to shut the fuck up and get on with life, but it is harder when you think they are prominent figures in whatever the hell you consider the audience or target of your software to be. Instead I decided that, would people tolerate such disgusting statements, they too could get fucked, and evidently they were not going to gain anything from a replicated object system which would only have distributed moderation and programming facilities anyway. And having to formalize what I was thinking made me feel worse about it.
Some months later (i.e. two weeks ago) I finished off the sequel article entitled Terminal boredom, which I also wrote as I feared that Netfarm would be considered too large to be useful. This fear nearly drove me to not publish in the European Lisp Symposium, as it seemed that the "conventional wisdom" that any client-side programming is necessarily a security hole would have me laughed out of the virtual room. But I was convinced by others to go on anyway, and that ended up not happening. However, writing Terminal boredom and researching lines-of-code counts made me realise that, really, the simplicity of less-is-less protocols is mostly marketing speak, and all of Netfarm (client, server, DHT, virtual machine, you name it) is only some 50% larger than a nice client for such a protocol.
So the fate of my work is not guaranteed to be failure.