TBH I think the distro model probably works really nicely for C-and-friends, from days of yore. When there was no package manager for C, resources like disk space were limited and the “ecosystem scale” of programming was a bit smaller. In this situation the benefits provided by distro-package-managers are large.
However, I think programming and the ecosystem is different these days and I am beginning to think the “single globally shared dylibs” model isn’t a good fit for many of the more recent programming languages and ecosystems.
We’ve seen it with the bcacheFS stuff and I think we’ll see it with more stuff in the future. I suspect there’s fundamental culture differences.
As mentioned elsewhere in the thread, I’d be interested in a distro model that split out “OS” and “app” level management and does not concern itself with “user”/“app” level dependencies, beyond maybe tracking the installed versions of apps, and little else.
However, I think programming and the ecosystem is different these days and I am beginning to think the “single globally shared dylibs” model isn’t a good fit for many of the more recent programming languages and ecosystems.
We’ve seen it with the bcacheFS stuff and I think we’ll see it with more stuff in the future. I suspect there’s fundamental culture differences.
As mentioned elsewhere in the thread, I’d be interested in a distro model that split out “OS” and “app” level management and does not concern itself with “user”/“app” level dependencies, beyond maybe tracking the installed versions of apps, and little else.