A significant thing happened to Hubski this week. rob05c completed the migration of our data from Arc's non-database to postgreSQL.
This was no small undertaking, which you may have followed on #devski.
rob05c did this on his own time, and we are in a much better place for it. Our ability to scale, to improve the app, and to offer a genuine API have been greatly expanded. In short, our previous architecture was untenable, and rob05c put us on a sustainable path.
I'll leave it to rob05c to discuss details, but know that from an architectural standpoint, Hubski is in a better place than it has ever been.
Thanks, rob05c. You get all the alfalfa.
The big site cost is paying for memory, because we currently load everything into memory. The SQL database doesn't directly fix that, but it makes it possible to load data as-needed. Once we actually do that, we'll be able to pay for hosting with less memory. It's not quite as easy as it sounds. Changing a data field to load from SQL where it's needed is easy. The problem is, the code currently assumes data is cheap, and will do things like load that field for every post ever, just to get it for a single post. That has to be fixed. But we'll get there.