2020-01-28 | Subject | Ecosystem Escape
I finished up the first pass on the sync routine for CB. It will allow simple collaboration. The scenario is a bunch of people in a room or online collaborating on documentation, but my implementation is primarily so I can work next to Sean in the other room on my articles and systems documentation.
The next problem is to get all of my Python code running on an Ubuntu 18.04 install. I am avoiding the Python package distribution ecosystem. The idea of a software ecosystem isn't that new. Perl had had/has it with CPAN and others. There are problems with relying on this. I cannot be assured that "pip install wxpython" will always work for my needs. Perhaps I don't have internet connectivity? It gets a bit weirder. I am currently compiling wxpython from source in another screen, and it is hung up on all of the sip compiles.
Sip is owned by Riverbank Computing. Sure, it is opensource. But the world does tend to go sideways a bit. Projects fork off or deprecate particular features. Companies (cough... Oracle... cough) change their stance. I remember trying to get a full Webkit browser running under QT, and I ran into some issues with license changes for the features I was using. My mission is particularly complicated, because I also test with MacPorts. My main point is that a snapshot in time, like a complete Ubuntu package repository and wxpython source code, is quite useful and facilitates autonomy. Autonomy... my-o-my we are painting ourselves in a corner with cloud in that regard. I am not so sure I will approve any project without an option to "bring it home". Dealing with some of the huge underpinnings of projects that are run by commercial companies, both for infrastructure and code, is another important consideration.
I could go on a bit about why the current software ecosystem is so fragile in general, but I don't really have to. All I have to do is say that my mission is to make it so that I can show people how to build an AI platform that generates documentation. I can show it end-to-end, from the initial disk image booting to running SPARQL queries. And, yes, AI is pushing it as a commonly understood name. Really, this will be more of an expert system. The inferences it will make are relatively simple. Now, when you make simple inferences on data flow models of an organization, this is miles above Visio, so... win. It is pretty similar, though, to the AI that is talked about so often these days.xernie