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
Hi, I use this library to get releases from GitHub, and I noticed something I want to get clarity about. I have a new release 0.4.0, but it's being reported as "not compatible" with the current version of 0.3.0. I double checked this behavior with the following, but you basically already have this as a test case here.
This seems surprising, because it means only patch versions can be incremented in a 0.y.z version and still be treated as compatible. Is this intentional? I'd expect minor versions to be allowed to increment and be compatible even with 0.y.z versions, similar to how it is compatible in 1.y.z versions.
Eg, if the logic was changed to this (and the single 0.2.0 -> 0.3.0 test case changed), this would make sense to me:
This is to mirror what cargo-update considers a compatible bump. For versions < 1.0.0, minor changes 0.2.0 -> 0.3.0 are considered breaking in the same way changes from 1.0.0 -> 2.0.0 are
This is to mirror what cargo-update considers a compatible bump. For versions < 1.0.0, minor changes 0.2.0 -> 0.3.0 are considered breaking in the same way changes from 1.0.0 -> 2.0.0 are
Interesting... Is that part of the semver spec, or something cargo-update is doing on their own separately?
Hi, I use this library to get releases from GitHub, and I noticed something I want to get clarity about. I have a new release
0.4.0
, but it's being reported as "not compatible" with the current version of0.3.0
. I double checked this behavior with the following, but you basically already have this as a test case here.I see that the bump compatibility logic is this:
This seems surprising, because it means only patch versions can be incremented in a
0.y.z
version and still be treated as compatible. Is this intentional? I'd expect minor versions to be allowed to increment and be compatible even with0.y.z
versions, similar to how it is compatible in1.y.z
versions.Eg, if the logic was changed to this (and the single 0.2.0 -> 0.3.0 test case changed), this would make sense to me:
Can you share more about why
0.y.z
versions don't treat minor increments as compatible bumps?The text was updated successfully, but these errors were encountered: