Hivemind sounds very complicated. I thought it simply was a second layer database, which doesn't sound like something that generates tons of bugs.
But then again, I am no dev.
You are viewing a single comment's thread from:
Hivemind sounds very complicated. I thought it simply was a second layer database, which doesn't sound like something that generates tons of bugs.
But then again, I am no dev.
Hivemind is responsible for responding to most of the API calls used by 2nd layer apps, so it's a fairly substantial sized service. And it has gained considerably more responsibilities as part of the eclipse upgrade. Calling it simply a database is probably the wrong way to look at it. It "has" a database, but then so do most web apps. The complexity is better measured in terms of the amount of functionality it provides.
I also don't want to imply that hivemind's design itself is "complicated" in a "difficult to understand" sense, it's not. It's just that there's a lot of queries and changes to the schema often requires those queries to be update. And those queries were also updated to make them faster. Between these two changes, it's a lot of changes to test.