While "full" sources for VSCode are available at https://github.com/microsoft/vscode the current ebuild actually installs a binary build made available by Microsoft. VSCode is known to include telemetry / tracking - even within the source available on github; and beyond that it's unclear what additional "secret sauce" might go into the binary. There is a project https://vscodium.com/ that tries to disable telemetry and tracking mechanisms against Microsoft servers, but that project immediately disclaims that this is not a 100% solution to stay clear of tracking and telemetry at all times. That project itself again primarily distributes a binary build. https://github.com/VSCodium/vscodium is to be understood as a build environment for https://github.com/microsoft/vscode rather than a fork. There is an Arch PKGBUILD available which looks like a good starting point to a possible future app-editors/vscodium "from source" build. So, although source code for VSCode is available (a feature somewhat heavily marketed by Microsoft), that does not mean it's free of "anti-features" from a FOSS perspective, which I presume Microsoft would like to be implied. To clearly signal both this and the fact that the ebuild requires resources to download the binary rather than building VSCode locally, I suggest this ebuild to be renamed to app-editors/vscode-bin, in line with chromium-bin, telegram-desktop-bin, signal-desktop-bin, openjdk-bin etc etc.
The current rough policy/guideline is that we usually elide the -bin if there's no way it can ever be built from source, fwiw.
I don't think there would be a truly build from sources vscode or vscodium in Gentoo. Setting up all the needed build environment for this will be very hard, and I'm pretty sure most users won't want to go for such lengths. There isn't a lot of advantages for that kind of application to build from scratch. If you see some kind of anti feature that can be turned off safely to be more FOSS, I'll gladly apply it (maybe behind USE flag, case by case decision). For *now* I think we will remain with the current naming schema.
