Users need clear mechanisms within Cloud IDEs to know the Java version, configure dependencies, build with specific versions, and identify/remove unwanted transitive dependencies.
Google’s internal IDE (Cider) was magical. I don’t miss a lot about Google (other than the free food and pampering LOL), but I do miss the IDE. Why would Google invest in a sizable team of SWEs building an internal IDE that they don’t make money on? Well, it doesn’t make money, but it saves money. A lot of it. With 120,000 SWEs at an average of $400k each (as per levelsfyi, SWE5 at Google hovers around $300k/yr plus a flat $100k/head estimate for perks, real-estate, overhead), my guess is Google spends $50B dollars per year (yes with a "b") paying their engineers. So yeah, having an IDE that can make these engineers even 10% faster at their job yields massive savings. What made it great was that it was a Cloud IDE. I didn’t have to install some crappy desktop app (I'm looking at you, IntelliJ and Visual Studio). I didn’t need to configure it. I didn’t need to install dependencies or worry about what version of Java I have in my laptop. I didn’t need to spend hours troubleshooting some cryptic error message triggered by me missing the small print somewhere in the setup documentation. I simply pointed Chrome at Cider and I had immediate access to all of google’s codebase. Already configured and magically built. Since development was on the cloud, so was build, of course, which was another nice bonus in terms of consistency and setup. And it was a Golden Path. Yes, you gave up some freedom: you couldn’t easily use VisualStudio or IntelliJ. Yes, there was a learning curve to learn a bespoke tool, which was absolutely useless the minute you left Google. But the benefits of being immediately productive were massive. When it comes to Engineering Productivity, Golden Path > Freedom, every time.