Deploy workflow to all PL Go repos

[marten-seemann]

Remaining steps:

• run go mod tidy in the copy workflow
• run gofmt -s in the copy workflow
• remove Travis and CircleCI config in the copy workflow (on first deployment)
• fix go vet, staticcheck and go test errors in all repos (see https://docs.google.com/spreadsheets/d/1Vwk-OUKXXNQkp4BkOoMndLylnJRW7a2c_mmvXu6jIj8/edit#gid=0 for the current status)
• add repos that are good to go to config.json

[BigLep]

Thanks for tracking this @marten-seemann .

A few things:

1. Am I correct to assume the "run x in the copy workflow" means making a change to https://github.com/protocol/.github/blob/master/.github/workflows/copy-workflow.yml ?
2. Who is going to "fix go vet, staticcheck and go test errors in all repos"? Will this needed to be done by the maintenance tributes? If so, we'll probably need to write up how many we want them to do per week, and what specifically we want them to do (e.g., checkout package, get go vet, statichcekc, and go test to pass, commit changes, update spreadsheeet, and add the repos to https://github.com/protocol/.github/blob/master/.github/workflows/config.json
3. Let's please add a task for checking in your script that generated https://docs.google.com/spreadsheets/d/1Vwk-OUKXXNQkp4BkOoMndLylnJRW7a2c_mmvXu6jIj8/edit#gid=0 (I don't have permissions to edit the issue)

[marten-seemann]

1. Yes. PR: run go mod tidy (and set Go 1.15) and gofmt -s in copy workflow #42. PR for disabling Travis / CircleCI coming soon.
2. I'll write up a document with instructions. Will post link here when done.
3. Even better, I created a draft PR: add the script used check the Go repo health #43. Draft because it depends on another PR that's not merged yet.

[marten-seemann]

Here's the manual how to contribute to our CI health: https://hackmd.io/@marten-seemann/BkVlbRMBO.
Any comments on the proposed workflow? @BigLep @Stebalien @mvdan

[mvdan]

Sounds good to me - I just left a nit comment.

I slightly worry that vet/staticcheck will get some pushback, though. I fully expect vet to be rejected on go-ipld-prime for example.

[BigLep]

Thanks @marten-seemann ! A few things:

1. I don't know the go world to comment on vet/staticcheck but if this is going to be decisive, lets cover this early. Without knowing more here, I would want to make sure we're not making things unnecessarily painful on ourselves here. If there is division about these tools in the go community, lets keep to the standard of "most" other "prominent" go project. I really want to make sure that getting tests to successfully run with each PR isn't lost in this conversation as that is the minimum bar I think we need to hit.
2. I made a couple of small tweaks to the document.
3. I linked to the document from https://github.com/orgs/ipfs/projects/12#card-58209321. The note currently says to timebox for 1 hour. Let me know if you think more time is needed.

Thanks!

[Stebalien]

1. go vet is standard staticcheck is pretty common and a nice minimal subset. Eventually, we could go with a subset of golangci-lint, but that linter is a bit too much sometimes.