Shuttleworth: 2 year cadence for major releases: some progress

In theory libraries/etc should define stable APIs so that this shouldn't
even be an issue.

However, I can see the value in having stable releases that "just work."

That said, I'm not sure I like the idea of linux becoming more of a
monoculture. I've already had to deal with FOSS projects that point
fingers when some distro ships their package and doesn't custom-tailor all
of its dependencies per their recommended settings. This could lead to
even less support for anybody who doesn't run the "standard" configuration.

The problem with standard configurations, is how does a new standard get
developed? If the standard is libfoo v2.3, then nobody will run libfoo
v2.4.1 or even 2.4.25. So, how do the developers of libfoo know if their
library changes will break 47 other applications? Suppose the powers that
be take a look at libfoo v 2.4.52, and it turns out that lots of stuff
breaks, so they stick with 2.3, and the libfoo devs don't want to fix it
because it turns out the problem is major design changes that happened
around v2.4.12. For that matter, what is the point of even working on a
project when nobody will use the code until it is "stable" and the code
will never be "stable" until people start using it.

For distros like Ubuntu or RHEL that strive for "just works" and whose
target is people who don't want to tinker with their systems,
Shuttleworth's proposal is probably a good thing. However, I'd hate to see
more cutting edge distros go this route. Ultimately, I think it would be
bad for everybody, including the stable distros.