#535128 unify v4&v6 ip route output

Package:
iproute2
Source:
iproute2
Description:
networking and traffic control tools
Submitter:
martin f krafft
Date:
2017-12-30 20:12:15 UTC
Severity:
wishlist
#535128#5
Date:
2009-06-29 22:04:41 UTC
From:
To:
Given that IPv6 is becoming more and more ubiquitous (2.6.30 kernels
in Debian have ipv6 compile in, it's no longer a module), it would
make sense to unify the output of 'ip route' in similar ways as the
output of 'ip address' already includes both address families.

A simple call to 'ip route' should include IPv6 routes, e.g.

  2001:a60:f069::/64 dev wifi  proto kernel  metric 256  expires 2592088sec mtu 1500 advmss 1440 hoplimit 0
  fe80::/64 dev wifi  proto kernel  metric 256  mtu 1500 advmss 1440 hoplimit 0
  192.168.20.0/24 dev wifi  proto kernel  scope link  src 192.168.20.102
  default6 via fe80::204:75ff:fecf:3ff6 dev wifi  proto kernel  metric 1024  expires 1727sec mtu 1500 advmss 1440 hoplimit 64
  default via 192.168.20.1 dev wifi

In general, there won't be overlaps, and IPv4 and IPv6
addresses/routes are sufficiently different. The only clash is with
'default', and I suggest that the IPv6 default route should be
labelled 'default6'.

If -4 or -6 are explicitly specified, the output should only contain
those address families. But if neither is specified, the two should
be merged.

#535128#14
Date:
2017-12-29 21:04:13 UTC
From:
To:
Control: tags -1 upstream wontfix
2592088sec mtu 1500 advmss 1440 hoplimit 0
hoplimit 0
192.168.20.102 
kernel  metric 1024  expires 1727sec mtu 1500 advmss 1440 hoplimit 64

Hi,

We shouldn't diverge from upstream in this kind of behaviour, so
marking this as wontfix (just 8 years later ;-) ).

Feel free to ask for this upstream at netdev@vger.kernel.org but I
strongly suspect such a change would be considered backward
incompatible, since it would probably break all the nasty screen-
scraping systems out there, so it's not likely to happen.

#535128#21
Date:
2017-12-30 08:52:27 UTC
From:
To:
also sprach Luca Boccassi <bluca@debian.org> [2017-12-29 22:04 +0100]:

Oh joy. Any tool should always check for a tty on stdout and if none
is found, it should randomly reformat the output unless
--machine-readable is given. You know, easier said than done… ;)

Thanks for following up, I guess I'll rest my case.