This is when I learned that there is also a long-running processes format that uses packet-line to communicate. Thus it doesn't belong in transport anymore, which was the previous 'best fit'. Since it's a serialization format, setting up its own crate makes quite some sense.
Showing the changes to git-protocol (3 of 27 files). Show all
Unified · Split
| @@ -15,7 +15,7 @@ doctest = false · 14 unmodified lines | ||
| 15 | 15 | serde1 = ["serde", "bstr/serde1"] |
| 16 | 16 | |
| 17 | 17 | [dependencies] |
| 18 | −git-transport = { version = "^0.1.0", path = "../git-transport" } | |
| 18 | +git-packetline = { version = "^0.1.0", path = "../git-packetline" } | |
| 19 | 19 | |
| 20 | 20 | quick-error = "2.0.0" |
| 21 | 21 | serde = { version = "1.0.114", optional = true, default-features = false, features = ["derive"]} |
| @@ -1,4 +1,4 @@ | ||
| 1 | −use git_transport::RemoteProgress; | |
| 1 | +use git_packetline::RemoteProgress; | |
| 2 | 2 | use nom::{ |
| 3 | 3 | bytes::complete::{tag, take_till, take_till1}, |
| 4 | 4 | combinator::{map_res, opt}, |
| @@ -1,7 +1,7 @@ | ||
| 1 | 1 | mod parse { |
| 2 | 2 | use bstr::ByteSlice; |
| 3 | + use git_packetline::RemoteProgress; | |
| 3 | 4 | use git_protocol::parse_remote_progress; |
| 4 | − use git_transport::RemoteProgress; | |
| 5 | 5 | |
| 6 | 6 | #[test] |
| 7 | 7 | fn a_message_we_dont_understand() { |