You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Note: I know that semver only applies to the "public API" of a crate so I'm not 100% sure if this qualifies as a violation of semver since I'm not sure how the public API is defined here.
In that PR @dtolnay mentioned that we'd just go for it with the change and reassess later if this happened to cause issues for anyone:
I was hopeful that somebody would look into what possibility of breakage this has for existing formats, but in the absence of anyone having been willing to do that, I think it will be all right to reassess based on any reported consequences.
I'm well and truly late to the party hahah but unfortunately this does cause some issues.
At a high level you can see the issue like this:
git clone git@github.com:aptos-labs/aptos-core.git
# Change serde version in Cargo.toml to 1.0.153
cd testsuite/generate-format
cargo run -- --corpus aptos
I'm still trying to figure out how exactly this change actually breaks our code, I just wanted to report in with an instance of a breakage. Once I know precisely what it changes for us I'll chime in again, though this looks pretty complicated. I believe the tldr is we have code that produces a yaml representation of generated code. We diff that yaml to make sure no incompatibilities arise. This code generation no longer works with 1.0.153 I believe because some of the other tooling we use (serde-reflection) breaks with this change to aliases + nesting.
If it turns out that this is part of the public API then I can make a minimally reproducible example to help debug.
The text was updated successfully, but these errors were encountered:
banool
changed the title
Semver incompatible change landed in 1.0.153
Semver incompatible change (?) landed in 1.0.153
Oct 11, 2023
Note: I know that semver only applies to the "public API" of a crate so I'm not 100% sure if this qualifies as a violation of semver since I'm not sure how the public API is defined here.
This PR was released as part of 1.0.153: #2387.
In that PR @dtolnay mentioned that we'd just go for it with the change and reassess later if this happened to cause issues for anyone:
I'm well and truly late to the party hahah but unfortunately this does cause some issues.
At a high level you can see the issue like this:
I'm still trying to figure out how exactly this change actually breaks our code, I just wanted to report in with an instance of a breakage. Once I know precisely what it changes for us I'll chime in again, though this looks pretty complicated. I believe the tldr is we have code that produces a yaml representation of generated code. We diff that yaml to make sure no incompatibilities arise. This code generation no longer works with 1.0.153 I believe because some of the other tooling we use (serde-reflection) breaks with this change to aliases + nesting.
If it turns out that this is part of the public API then I can make a minimally reproducible example to help debug.
The text was updated successfully, but these errors were encountered: