Remove byteorder #15
Remove byteorder #15
Conversation
There was a problem hiding this comment.
Thanks so much!
So normally, I'd be happy to take the Rust 2018 edition changes, but I would like to keep the diff here as small as possible to minimize conflicts with my ongoing rewrite branch. So to that end, could you please:
- Drop the Rust 2018 migration commit.
- For the byteorder removal change, minimize the changes such that the code still uses the full endian name, e.g.,
NativeEndianinstead ofNE.
And I think that will make the diff here pretty small.
|
(Note that the rewrite already migrates to Rust 2018.) |
|
Sounds good! |
workingjubilee
force-pushed
the
remove-byteorder
branch
from
2e0564b
to
ebc8aa0
Compare
|
The last commit now just notes what was already functionally declared by the ci change and includes the toolchain file. |
| @@ -0,0 +1,2 @@ | |||
| [toolchain] | |||
| channel = "1.40.0" | |||
There was a problem hiding this comment.
Hmmm... I've never used a rust-toolchain file. Did you mean rust-toolchain.toml? In any case, I don't think that does what I want. I don't want to build and develop on this version, I just want to have it be supported. I think CI is good enough for that. Otherwise, I want to use whatever my default toolchain is. (Usually latest nightly.)
This takes #15 and shrinks its diff considerably. Namely, we match the byteorder API a bit more exactly, remove all changes to dense/sparse/state_id and keep the minimum things needed in lib.rs and Cargo.toml. While improvements are in general welcome, all of this machinery is gone in the rewrite. The only point of this change is to satisfy a request to remove a dependency, of which I am sympathetic to. Closes #15
This takes #15 and shrinks its diff considerably. Namely, we match the byteorder API a bit more exactly, remove all changes to dense/sparse/state_id and keep the minimum things needed in lib.rs and Cargo.toml. While improvements are in general welcome, all of this machinery is gone in the rewrite. The only point of this change is to satisfy a request to remove a dependency, of which I am sympathetic to. Minimizing the diff is important for sanely rebasing my rewrite branch on top of this. Otherwise, removing this dependency probably becomes more annoying than it's worth. Closes #15
|
I opened #16 to supplant this one. Mostly I just really wanted to get the diff down to as small as possible. (Or at least, localized to a new file that I know I can just drop when I rebase my rewrite on top of this.) |
This takes #15 and shrinks its diff considerably. Namely, we match the byteorder API a bit more exactly, remove all changes to dense/sparse/state_id and keep the minimum things needed in lib.rs and Cargo.toml. While improvements are in general welcome, all of this machinery is gone in the rewrite. The only point of this change is to satisfy a request to remove a dependency, of which I am sympathetic to. Minimizing the diff is important for sanely rebasing my rewrite branch on top of this. Otherwise, removing this dependency probably becomes more annoying than it's worth. Closes #15
This takes #15 and shrinks its diff considerably. Namely, we match the byteorder API a bit more exactly, remove all changes to dense/sparse/state_id and keep the minimum things needed in lib.rs and Cargo.toml. While improvements are in general welcome, all of this machinery is gone in the rewrite. The only point of this change is to satisfy a request to remove a dependency, of which I am sympathetic to. Minimizing the diff is important for sanely rebasing my rewrite branch on top of this. Otherwise, removing this dependency probably becomes more annoying than it's worth. Closes #15
|
ahhh, sorry! I misunderstood. |
This PR removes byteorder, formally raises the MSRV to 1.40.0 (according to a run of cargo msrv), and adds a rust-toolchain file in order to control MSRV drift.
Closes #12.