4 ms·
That is inaccurate. In short, Russ Cox has been bashing cargo's approach of preferring newer versions and lockfiles and a SAT solver (or really an approximation
by TheDong 9y ago
That is inaccurate. In short, Russ Cox has been bashing cargo's approach of preferring newer versions and lockfiles and a SAT solver (or really an approximation of one).
Apparently making the dependency-manager's code simpler is a goal worth removing developer's ability to be expressive and removing security updates.
- Conan_Kudo 9y agoI'd be surprised if the Go developers ever care to do this right. Google doesn't use any of this, and all their projects do extensive vendoring with heavy modifications (effectively forking them). Personally, I despise their approach to developing software, and on principle, I won't really develop software in Go because their attitude permeates through language and software development mechanisms.
- tcheard 9y ago> Apparently making the dependency-manager's code simpler is a goal That isn't the goal at all. The goal of Minimal Version Selection is to ensure that the version selected is closest to that which was tested by the app and its dependencies. > worth removing developer's ability to be expressive and removing security updates. And that isn't true either. It allows the developer of both the app and its dependencies to express the version they have tested against, and still allow the developer to receive security updates by explicitly increasing the version, allowing them to test that the update doesn't create problems while doing so. It allows for version pinning without the need for a lock file.