Also, like where else would the indices live? They are in memory to make going to definition, source, analysis etc fast. It's not even using java objects for these, so a feature-equivalent solution would be similarly "fat".
I don’t know, using Vim and Exuberant Ctags can give you a significant part of these features for a very lower footprint. But to be clear, it’s not a ready-to-go path on parity with DX that IntelliJ provides.
Hackers are often exceptionally good at optimizing the hidden parts no one give a shit until everything crumble under systemic slowness, but they often also are very bad at designing even a decent UX for everything else regarding in-your-face aspects.
>I don’t know, using Vim and Exuberant Ctags can give you a significant part of these features for a very lower footprint.
100% true.
Also, the problem I have with these magic thingies is that I don't know what to do if the magic stops working. With tools that just does one thing, I can easily figure out the point of failure and fix it.
When I worked as a Scala developer years ago, I still used JetBrains IDEs for some things but I did most of my work in Emacs using ENSIME and ggtags/xref as a fallback. I ended up using ggtags elsewhere as well, and was pretty impressed with both how far it got me and how fast it was.
Hackers are often exceptionally good at optimizing the hidden parts no one give a shit until everything crumble under systemic slowness, but they often also are very bad at designing even a decent UX for everything else regarding in-your-face aspects.