The moving target of OS support

[Christopher James Halse Rogers]

On Wed, Aug 12, 2015 at 2:08 AM, Kevin Gunn <kevin.gunn at canonical.com> 
wrote:
> I tend to agree, although i am curious to hear what others think.
> My hope would be that we'd be balanced about adopting "new language 
> variants and dependencies" - if we have reasons to do so, then do 
> those outweigh stagnating for the sake of being able to build on 
> older ubuntu stables?

I don't think “has to build on latest LTS packages” is a reasonable 
requirement. GNOME doesn't support that; the X server doesn't support 
that; Mesa doesn't support that.

Developers are used to needing updating dependencies. So much so that 
there are very good tools for managing this, such as jhbuild.

Now, requiring a new *compiler* is a more disruptive requirement. Even 
here the foundations team maintains a PPA with new toolchains. I also 
don't see us requiring new compiler features in the immediate future; 
C++17 support isn't going to be reasonable for some time :).

We should be mindful of the cost of adding or updating a dependency, 
but not hesitate to do it if doing so makes our jobs easier.