Re: 9.0 is getting old...

[Greg Troxel <gdt%lexort.com@localhost>]

"J. Lewis Muir" <jlmuir%imca-cat.org@localhost> writes:

> On 06/07, Greg Troxel wrote:
>>   I see pkgsrc as supporting 9.x, not 9.0
>
> Not saying you're wrong in thinking that pkgsrc should be supporting
> and targeting 9.x instead of 9.0, but this seems different from what
> I thought was being targeted, so it would be great if that could be
> documented somewhere.

Well, pkgsrc is the source code, and we try to have packages build
everywhere, but people fix what they fix.

Here, I was sloppy in wording, and this is "binary packages built by
TNF" which has two subcategories, packages built on TNF-owned machines
(x86, arm) and packages built by individuals (lots of retro and
semi-retro arches).

I do have a bias that anyone running a branch of NetBSD should track
that branch, and that running 9.0 is no longer a reasonable thing to do.
So I care much more about a good outcome for 9.3/9.4 than 9.0.

We do tend to use the .0 for the builders, unless it is trouble.

>>   I do not expect binary packages built on 9.3 to be troubled on 9.0.
>>   The theory is that we have binary compatibility.
>
> I don't understand this statement.  Say a new function were added to
> a 9.3 userland library that didn't exist in 9.0, and a pkgsrc package
> that used that new function were built against a 9.3 userland, then
> the resulting pkgsrc binary or library wouldn't work on a 9.0 userland
> because the function that was added to 9.3 doesn't exist in 9.0.

Technically you are right if there was an ABI addition.  But we try not
to do that either; it's mostly fixes and new drivers/devices in the
kernel.

There is no truly great answer here; every approach is going to bother
somebody.