Batis - XDG-based packaging for Linux desktop apps

[Jerome Leclanche]

On 20 November 2015 at 22:54, Thomas Kluyver <thomas at kluyver.me.uk> wrote:
>
> Also, I'm not sure the Linux desktop has much of a future unless the
> situation improves sooner. The number of Linux desktop users I know is
> dwindling as they convert to Mac. Programming conferences, even where
> open source software plays a big part, are now a sea of Macbooks. We
> need bigger plans like sandboxing technology, but we also need something
> to move the status quo forwards.

I get where you're getting at but yet-another-packaging-format is not
the solution...
You can write wrapper over wrapper over wrapper and all you end up
with is a thousand wrappers. You didn't fix the underlying issue.
Every format sucking in its own right means there's no clarity in what
you should distribute as an app developer. If you're not intimately
familiar with linux, your problem is "Hey, should I distribute debs?
rpms? tarballs? can I not distribute binaries? do I need to let the
distro handle my stuff? what's that pacman they're talking about?" --
adding another item into that particular mix is simply not useful.

This is a problem that fixes itself the moment distros start
converging towards a common format. Part of the issue is technical for
sure, but the technical problems don't reveal themselves until after
you know what the political solutions are.
So deb, rpm and tar.*z are really all the same format, with differing
metadata on top. This is the big item to outsiders, and it's the
easiest item to converge on. Where it gets hard is "agreeing" on the
resulting structure, the various etc/usr/whatever prefixes and what
not. Every distro has, in their own eyes, equal claim to their own way
of doing things - you can't just say "Debian is doing it right. Now
we're debian."... and whatever you do decide means a massive migration
for everyone else *and* you didn't really fix the problem. Smaller
distros doing-their-thing could become big with just a little
corporate backing and you're back to square 1.

The sandboxing drive we're seeing (aka the android model) moves that
problem to a different layer. You're now packaging for the layer,
similarly to distro maintainers that are packaging for their layer
(the distro). If you want to fix packaging, this is the only
non-whackamole route. But it's not enough to do that in a vacuum, you
also have to have a global vision of each distribution's problems,
solutions and reasonings which is a far bigger task, hence why the
technical aspect is insignificant in comparison.
What we're seeing with Limba (what Jasper linked) for example is a
saner approach. It's still just one piece of the puzzle, but it
understands that 1. The larger problem is metadata and 2. A solution
in a vacuum is not a solution.


J. Leclanche