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
i note that release 0.8 appears to be pretty stable for string_cache -- any chance we might consider the API to be stable, and signifying that with a semver 1.0.0 release?
Cargo treats 0.8, 0.9, 0.10 etc. the same as 1.0, 2.0, 3.0, etc. The the difference is only symbolic. Is this alone worth putting the ecosystem through another breaking change?
There is a difference between 0.8, 0.9, ... and 1.0, 2.0: cargo will automatically update 1.0 to 1.1, but will refuse to update 0.8 to 0.9. This makes it harder for users to upgrade and increases the number of people still on old versions.
Is this alone worth putting the ecosystem through another breaking change?
There is a difference between 0.8, 0.9, ... and 1.0, 2.0: cargo will automatically update 1.0 to 1.1, but will refuse to update 0.8 to 0.9.
… yes? And it’ll automatically update 0.9.1 to 0.9.2 and refuse to update 2.1.6 to 3.0.0. So?
This makes it harder for users to upgrade and increases the number of people still on old versions.
Difficulty doesn’t change if we do 0.8 then 0.9 then 0.10, compared to 1.0 then 2.0 then 3.0.
Is this alone worth putting the ecosystem through another breaking change?
I'm not sure what you mean by this.
Making a new version that Cargo won’t automatically upgrade to only makes worse the problem you mentioned of various users depending on mutually-incompatible versions of the library. This can cause larger binaries at best when an app gets multiple copies of the library, or type errors when trying to use different version together in public APIs.
I’d prefer to have a better reason to do this rather than only having a prettier-looking version number.
I'd like to ask for a 0.8.1 or 0.9 release to publish #235, which makes the serde dependency optional but enabled-by-default. This should be 0.9 because someone could have used default-features = false while relying on serde support (even though string_cache 0.8.0 doesn't expose any other features).
@SimonSapin please consider a 0.8.1 or 0.9 release, I have something that is seriously bogged down by a transitive serde dependency, which you already fixed with #235 11 Month ago, but haven't released. Is there any technical reason this can't be done?
https://crates.io/crates/string-cache/0.8.1 is up and includes #235. Breakage is indeed possible with default-features = false, but I don’t know if there’s consensus that this should be covered by (cargo-modified) semantic versioning. And if you have an application with multiple dependencies that depend on string-cache, migrating everything to 0.9 would create busywork.
@dkg and @Ygg01, what do you mean by stable? I feel it could mean either "causes crashes" or "changes a lot", but neither really applies to string-cache in my opinion. 1.0 v.s. 0.x is all about communicating a certain image to potential users. If there is no other breaking change, image alone does not seem worth releasing a semver-incompatible version to me.
Activity
SimonSapin commentedon Jan 6, 2020
Cargo treats 0.8, 0.9, 0.10 etc. the same as 1.0, 2.0, 3.0, etc. The the difference is only symbolic. Is this alone worth putting the ecosystem through another breaking change?
jyn514 commentedon Feb 9, 2020
There is a difference between 0.8, 0.9, ... and 1.0, 2.0: cargo will automatically update 1.0 to 1.1, but will refuse to update 0.8 to 0.9. This makes it harder for users to upgrade and increases the number of people still on old versions.
I'm not sure what you mean by this.
SimonSapin commentedon Feb 9, 2020
… yes? And it’ll automatically update 0.9.1 to 0.9.2 and refuse to update 2.1.6 to 3.0.0. So?
Difficulty doesn’t change if we do 0.8 then 0.9 then 0.10, compared to 1.0 then 2.0 then 3.0.
Making a new version that Cargo won’t automatically upgrade to only makes worse the problem you mentioned of various users depending on mutually-incompatible versions of the library. This can cause larger binaries at best when an app gets multiple copies of the library, or type errors when trying to use different version together in public APIs.
I’d prefer to have a better reason to do this rather than only having a prettier-looking version number.
robojumper commentedon Oct 15, 2020
I'd like to ask for a
0.8.1or0.9release to publish #235, which makes theserdedependency optional but enabled-by-default. This should be0.9because someone could have useddefault-features = falsewhile relying onserdesupport (even thoughstring_cache 0.8.0doesn't expose any other features).Thanks in advance!
mvtec-bergdolll commentedon Nov 10, 2020
@SimonSapin please consider a
0.8.1or0.9release, I have something that is seriously bogged down by a transitive serde dependency, which you already fixed with #235 11 Month ago, but haven't released. Is there any technical reason this can't be done?Ygg01 commentedon Nov 18, 2020
@SimonSapin I second bergdolll and robojumper. Release a
0.9would be enough for now. With stabilization a major milestone.SimonSapin commentedon Nov 19, 2020
https://crates.io/crates/string-cache/0.8.1 is up and includes #235. Breakage is indeed possible with
default-features = false, but I don’t know if there’s consensus that this should be covered by (cargo-modified) semantic versioning. And if you have an application with multiple dependencies that depend onstring-cache, migrating everything to 0.9 would create busywork.@dkg and @Ygg01, what do you mean by stable? I feel it could mean either "causes crashes" or "changes a lot", but neither really applies to
string-cachein my opinion. 1.0 v.s. 0.x is all about communicating a certain image to potential users. If there is no other breaking change, image alone does not seem worth releasing a semver-incompatible version to me.Ygg01 commentedon Nov 23, 2020
Seems good enough for me.