[jmull]

A perfectly valid evolution strategy is to add new APIs while keeping the old (which you may or may not choose to deprecate). That is, add the new APIs to 3.x (but probably call them something other than v4.

Now, that's effectively what 3.25 is. But there are some problematic extras... the semantics of the semantic versioning is muddled. That means confusion, which means some people will spin their wheels after not initially grokking the situation correctly. Also, there seems to be an implication there could be a strong deprecation of v3 APIs coming. That is, you have to wonder how long the window for incremental migration will remain open.

To be frank, I wouldn't touch zod on any project I hoped will be around for a while. Predicting the future is an uncertain business, but we can look at the past. In a few years, I think it's reasonable to guess zod v4 will be getting the same treatment zod v3 is getting now.

Not that I think you ought to do anything different. I probably wouldn't want to maintain some old API I came up with years ago indefinitely either (not without a decent support contract, that is).