Linux software sources are in a weird spot between users and developers.
I, as a developer, organize sources in a way that make it easy to work for another developer. My software will never be compiled by any user. All my users use build artifacts.
I might consider adding autogenerated code, but only when I'm like 99% sure that this code won't ever change. For example that's the case for integration with many organizations where WSDLs are agreed upon once and then never touched. Having Java sources regenerated every build just adds few seconds to every build time without noticeable advantages.
The fact that some Linux users prefer to build software from the sources and at the same time do not want to install necessary build tools is a bit strange situation.
May be containers should be better utilized for this workflow. Like developer supplies Dockerfile which builds a software and then copies it to some directory. You're running `docker build .` and they copying binary files from the container to the host.
I, as a developer, organize sources in a way that make it easy to work for another developer. My software will never be compiled by any user. All my users use build artifacts.
I might consider adding autogenerated code, but only when I'm like 99% sure that this code won't ever change. For example that's the case for integration with many organizations where WSDLs are agreed upon once and then never touched. Having Java sources regenerated every build just adds few seconds to every build time without noticeable advantages.
The fact that some Linux users prefer to build software from the sources and at the same time do not want to install necessary build tools is a bit strange situation.
May be containers should be better utilized for this workflow. Like developer supplies Dockerfile which builds a software and then copies it to some directory. You're running `docker build .` and they copying binary files from the container to the host.