I haven't ever maintained a blog, but since Linux Journal folded and I no longer work at Purism, I miss having an outlet for longer-form thoughts.
Over the last five years or so, I've observed a Renaissance in innovation on the Linux desktop that hearkens back to the Golden Era of the desktop from the mid-1990s into the aughts.
In this (pretty long, likely controversial) post, I talk through the Golden Era, Dark Ages, Renaissance, and what's next.
@kyle It's kind of funny how you state that apt helped solve dependency management issues, but a new hope is being presented in flat packaging, what I would call extra-distro dependency avoidance at a storage cost.
@lwriemen At the time apt came about, we did have a way to essentially "vendor" dependencies with software via static linking, but storage was too precious so dynamic libraries were preferred. The hope that containers presents is in no longer having to factor in system software dependencies in your code (dependencies that are often out of date). It comes at some storage cost, but for desktop systems storage isn't nearly at the kind of premium anymore.
@lwriemen Space savings is a tougher call. Even if you only have a single VM for everything, you still need that space for that base image. Base images inside flatpaks are likely smaller than a base Qubes Debian/Fedora VM.
If the base VM image doesn’t have the system libraries a particular desktop app needs, and other desktop apps in a different Qube uses some of the same libraries, I would think both of those layers on top of the base VM would have dupe packages in them. So a toss-up there.
@lwriemen Qubes definitely has a security advantage, because it is using hardware virtualization features to isolate software from each other, and with containers you are relying on software controls enforced by the kernel (if enabled by tools like bubblewrap within the container to begin with, some Flatpaks are pretty locked down, other’s less so).