[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20241028174343.0bc36cea@hermes.local>
Date: Mon, 28 Oct 2024 17:43:43 -0700
From: Stephen Hemminger <stephen@...workplumber.org>
To: Jakub Kicinski <kuba@...nel.org>
Cc: David Ahern <dsahern@...nel.org>, Paolo Abeni <pabeni@...hat.com>, Joe
Damato <jdamato@...tly.com>, "netdev@...r.kernel.org"
<netdev@...r.kernel.org>
Subject: Re: yaml gen NL families support in iproute2?
On Mon, 28 Oct 2024 17:06:47 -0700
Jakub Kicinski <kuba@...nel.org> wrote:
> On Mon, 28 Oct 2024 16:40:55 -0700 Stephen Hemminger wrote:
> > > I can only find this thread now:
> > > https://lore.kernel.org/all/20240302193607.36d7a015@kernel.org/
> > > Could be a misunderstanding, but either way, documenting an existing
> > > tool seems like strictly less work than recreating it from scratch.
> >
> > Is the toolset willing to maintain the backward compatibility guarantees
> > that iproute2 has now? Bpf support was an example of how not to do it.
>
> The specs are UAPI.
>
> The Python and CLI tooling are a very thin layer of code basically
> converting between JSON and netlink using the specs, so by virtue
> of specs being UAPI they should be fully backward compatible.
>
> The C library is intended to be fully backward compatible, but right
> now only supports static linking.
Right but now you can update kernel and CLI tooling, and not worry about
running on older kernels.
Powered by blists - more mailing lists