Hacker News Reader

To be fair, Go has quite successfully (IMO) not implemented features, waited for the community to come up with several different implementations, and then implemented that feature in a way that they like, without fracturing the ecosystem. Examples that come to mind are `//go:embed` and dependency management via `go mod`.

View on HN

eh, go mod isn't the great example...the community was pretty fractured before that. I'd personally adopted dep right before they announced the intention to build modules...so that was kind of frustrating.

in the long run i think it'll consolidate the community, but there's definitely a long tail to that timeline.

But were Go modules a better solution in that space? Should we have stuck with a less better approach (not casting a judgement on dep necessarily) just because it was an earlier attempt to deal with the same problem?

so far in my experience it's a bit meh on both fronts. modules got rid of GOPATH, which is kind of nice. i personally never cared much about the whole SAT vs ...whatever algorithm rsc implemented (I forget the details at this point, which may be an indicator of how important it is the the user base...or I'm just ignorant of the benefit's I'm getting/missing)

having it be incorporated into the project is ultimately the real win, because it gives a target to consolidate round.

Reply on news.ycombinator.com