Major releases and breaking changes

[phil294]

I was always under the impression that Deno would be backwards-compatible eternally, based on

As of Deno 1.0.0, the Deno namespace APIs are stable. That means we will strive to make code working under 1.0.0 continue to work in future versions. - Deno Manual, Stability / Release Schedule

But this is obviously not the case, as there are several breaking changes listed above. Could you release some information on how often major breaking version changes are expected in Deno? This seems like a fairly important question to consider, especially in enterprise context. For example, its strong backwards compatibility is arguably one of the core reasons Java has succeeded in staying relevant over the years. Personally I'm a big fan of Deno but would not really consider it in any large-scale project things keep changing so much.

In other words, don't you think it would be enough (at least regarding Deno namespace changes) to deprecate things, but not break them?

It's already happening, unfortunately, as a lot of Deno code found on the internet does in fact not work anymore (pre 1.0, especially). Often, imports in code samples are dead (removed from std, which itself is also still not stable). A phenomenon quite familiar from web browsing, but in programming languages, a rather unique aspect of Deno.

Anyway, what's the philosophy like for future breaking changes, especially long-term? I was unable to find any information on this

(Not taking into account TS version bumps)

(migrated from #12110, where you can also find answers on this by @vwkd and @GJZwiers