Contributed by pitrh on from the all sixes and routes dept.
I'm not entirely sure if mg(1) has dedicated maintainers, but if there are it's probably lum@, jasper@ and me. When u2k15 was first discussed at c2k15 in Calgary someone said: "Florian, will you come? We need utf8 for mg!" Me: "For sure I will come, I will build myself an IT throne and mock anybody who touches mg(1)'s input handling."
But seriously, I do agree that mg(1) needs to understand utf8, but I'm not competent (at this time) to make this happen. I try to follow along what's going on with utf8, but for u2k15 I joined the network hackers.
phessler@ and claudio@ introduced the concept of an emergency hackathon at g2k14, and I went with that option: Fly in on Friday evening and leave at stupid o'clock on Monday. It became blatantly obvious that my traveling would be quite different from my c2k15 experience when I asked for a beer on the flight and got handed a Heineken...
At the hackathon I continued poking ping(8) and ping6(8) to see what would stick to eventually merge the two into one binary. This led to the -g option in ping6(8). Which then led to the removal of IPV6_NEXTHOP support from the kernel to the great joy of mpi@. Source routing isn't considered to be a good idea these days, or as claudio@ pointed out, it wasn't a good idea 20 years ago.
mpi@ then pointed out that linkmtu and maxmtu from struct nd_ifinfo and more importantly the IN6_LINKMTU macro were getting in his way. Handling the mtu in the kernel on a received router advertisement message doesn't seem to be the right way to handle this. I think this was the first hackathon where I commited a net minus...
˙uǝddɐɥ uoɥʇɐʞɔɐɥ sıɥʇ ƃuıʞɐɯ ɹoɟ ǝsʃǝ ʎpoqʎɹǝʌǝ puɐ uıʃɹǝq-uı '@ǝʍn '@dsʇs oʇ sʞuɐɥʇ
Thanks for the deep probe in the code and the report, Florian!
(Comments are closed)