Bug#845304: transition: libxtables12
Hi!
On Tue, 2016-11-22 at 10:49:11 +0100, Arturo Borrero Gonzalez wrote:
> On 22 November 2016 at 10:39, Emilio Pozuelo Monfort <pochu@debian.org> wrote:
> > To the maintainer: Why bump to a snapshot at this point in the cycle? Have the
> > rdeps been build-tested against the new libxtables? Are you aware we are in the
> > transition freeze and this is a transition?
> I bumped to a snapshot release because several people asked me to do so.
I was one of those. The changes in the snapshot are requiered for the
various translation tools, to convert from the legacy iptables to
nftables command-line syntax, or to be able to inject rules into
the nftables using iptables commands and syntax, otherwise the tools
do not see each others rulesets. I'm sorry this has caused grief for
Arturo and the Release Team. :(
We had an in-person chat several days ago with the principal
iptables/nftables developer and we mentioned that having an actual
release would be helpful, and he said that he might be doing that
soonish?
I can understand why you'd prefer a revert at this point in time, although
I think it's worth considering that given that this transition has already
been started (even though, unfortunately, via an accident), that it
involves very few packages, that it will have a positive effect on people
wanting to migrate to use nftables in Stretch, that the upload was done
in good faith, and reverting might be messy as well, perhaps it actually
it's worth letting it in?
OTOH, the conversion is usually a one-off thing, so I guess this could be
handled via backports or on another system with more up-to-date userland.
On Tue, 2016-11-22 at 12:05:35 +0000, Simon McVittie wrote:
> On Tue, 22 Nov 2016 at 10:49:11 +0100, Arturo Borrero Gonzalez wrote:
> > I missed the point that libxtables (which is in src:iptables) added a
> > few new symbols that triggers transitions.
>
> Wait, *added* a few new symbols?
It also changes the «struct xtables_match», which is passed and returned
as part of the public API. So the SOVERSION bump seems warranted to me
(upstream commit 7a0992da44cfb6cab0ccd1beadcf326df8773552).
Thanks,
Guillem
Reply to: