<cr1901_modern>
might be an interesting patch though
<cr1901_modern>
Incidentally, TIL that Record()'s constructor doesn't add a default direction if you pass (name, size) layout elements
<GitHub168>
[smoltcp] jackpot51 commented on issue #66: Yep, will do https://git.io/vFcpv
<GitHub164>
[smoltcp] jackpot51 commented on issue #66: @whitequark it is done https://git.io/vFcpF
<GitHub21>
[smoltcp] whitequark commented on issue #66: Any chance you can write a test for this? https://git.io/vFchJ
<GitHub62>
[smoltcp] jackpot51 commented on issue #66: Sure! https://git.io/vFchY
<GitHub67>
[smoltcp] jackpot51 commented on issue #66: @whitequark I do not see any current tests for UDP in the ethernet interface, am I mistaken? https://git.io/vFchC
<GitHub102>
[smoltcp] dlrobertson commented on pull request #69 3a0d9f4: the `timestamp_ident` and `timestamp_seq_no` are exact copies of `echo_ident` and `echo_seq_no`, but I added them for consistency/readability. https://git.io/vFCe7
<GitHub135>
[smoltcp] dlrobertson commented on pull request #69 3a0d9f4: Added `TimestampRequest` and `Timestamp` reply. They're not essential to this PR. The main reason for adding them was for the test case in `iface::ethernet::test::test_icmpv4_socket` https://git.io/vFCe9
<GitHub193>
[smoltcp] dlrobertson commented on pull request #69 3a0d9f4: Updated this to be more consistent with the other packet enums https://git.io/vFCeH
<GitHub127>
[smoltcp] dlrobertson commented on pull request #69 3a0d9f4: This is what I really don't like about this implementation. I originally wanted to stay away from using a `bind` call, but now as a result if you need to use a `recv` as the `bind` in a way. https://git.io/vFCeQ
<GitHub158>
[smoltcp] whitequark commented on pull request #69 3a0d9f4: Ugh. Just after I migrated `socket-*` features to `proto-*` I think we'll need to go back. https://git.io/vFCvf
<GitHub84>
[smoltcp] whitequark commented on pull request #69 3a0d9f4: This match statement *begs* being put in a macro. https://git.io/vFCvJ
<GitHub33>
[smoltcp] whitequark commented on pull request #69 3a0d9f4: That request is deprecated, no? I try to avoid adding deprecated things to `wire` unless they're used in practice. https://git.io/vFCvk
<GitHub62>
[smoltcp] whitequark commented on pull request #69 3a0d9f4: Since this socket works with raw packets the TTL is already in there. https://git.io/vFCvT
<GitHub70>
[smoltcp] whitequark commented on pull request #69 3a0d9f4: I think this will need to be reworked to have a proper `bind` call, at least if we want to support downstreams like redox which present a POSIX-like interface. Take a look at this: https://www.planet-lab.org/raw_sockets/api_icmp.html. https://git.io/vFCvI
<GitHub111>
[smoltcp] whitequark commented on pull request #69 3a0d9f4: I think that ICMP sockets could easily serve both IPv4 and IPv6. After all, the first nibble of the header already tells you which packet it is. https://git.io/vFCvU
<GitHub194>
[smoltcp] dlrobertson commented on pull request #69 3a0d9f4: Np. Should I add that in this PR? https://git.io/vFCvi
mumptai_ has joined #m-labs
<GitHub68>
[smoltcp] dlrobertson commented on pull request #69 3a0d9f4: Yeah, it's not really used. It's really only used when echo requests are blocked by a firewall. I'll just manually write out the bytes for the test in the interface tests. https://git.io/vFCv9
<GitHub111>
[smoltcp] dlrobertson commented on pull request #69 3a0d9f4: We still use `IpRepr::emit`, so we still need to populate the ttl of the repr [here](https://github.com/dlrobertson/smoltcp/blob/icmpv4_socket/src/socket/icmpv4.rs#L209-L224) right? It is a "raw packet" but I made the assumption that the raw data passed to `send` would not include the IP header. That may be an error in my implementation. https://git.io/vFCf3
<GitHub8>
[smoltcp] dlrobertson commented on pull request #69 3a0d9f4: That is a super helpful doc https://git.io/vFCfZ
<GitHub103>
[smoltcp] dlrobertson commented on pull request #69 3a0d9f4: Id be interested to see how the type based `binds` are implemented for IPv6. AFAIK we'd need to support the Fragment header to get an identification number, but the id is 32 bits in IPv6. I'll have to dig through linux and/or some docs. https://git.io/vFCf8
<GitHub22>
[smoltcp] whitequark commented on pull request #69 3a0d9f4: Why not use an echo request? https://git.io/vFCJu
<GitHub114>
[smoltcp] whitequark commented on pull request #69 3a0d9f4: Don't bother with IPv6 for now, its headers are fairly broken and we'll need to aggressively compress them into Reprs. https://git.io/vFCJz
stekern has quit [Remote host closed the connection]
_rht has quit [Quit: Connection closed for inactivity]
<sb0>
whitequark, mostly looking for optical/laser parts now. vacuum is less of a pain in the ass.
<sb0>
vapor cells, wavemeter, misc parts like mirrors, corner cubes, mounts, etc.
<GitHub36>
[artiq] sbourdeauducq commented on issue #848: He wants to run multiple masters one one host. You can have multiple hostnames mapped to one IP, and different ports open for each master, but then there can be confusion if you connect with an unplanned hostname/port combination.... https://github.com/m-labs/artiq/issues/848#issuecomment-341896553
<GitHub25>
[smoltcp] dlrobertson commented on pull request #69 ceeed1c: Only comment I haven't fixed yet. Thinking about how to best accomplish this since some branches need `ChecksumCaps` etc. https://git.io/vFCXT