Awesome-go: maybe stop relying on "golint"

[kevinburke]

There are a lot of problems with it and situations where it flags things that are not actually problems. This is confusing and discouraging. It also doesn't have a maintainer.

Just to name a few from the issues list.

https://github.com/golang/lint/issues/354
https://github.com/golang/lint/issues/350
https://github.com/golang/lint/issues/286

Along with others I've run into, for example on https://github.com/avelino/awesome-go/pull/1698:

error: ExampleLoad refers to unknown identifier: Load (vet)

This is confusing, because there is an identifier named Load in that package, and godoc handles it with no problems; it looks like there's an error in the script generating the report card recommendation.

[kirillDanshin]

error: ExampleLoad refers to unknown identifier: Load (vet)

it has nothing to do with golint. it's the go tool vet output. if you don't get it via go tool vet --all, you may want to enable more rules or update the Go to latest stable.
there is nacl.Load(), but you're trying to write an example in nacl_test, not the nacl where Load() is defined. so that's not a vet's bug, it's yours. to omit this error you typically use something like Example_Load.

golint edge cases can be fixed, but I even didn't seen before real-life use of that like piece of ~sh~ code, that can trigger that bugs.
even more, I think golang/lint#286 can be ignored 'cause that's not the way you should do that.

thanks for feedback, we appreciate it, but this time it's not relevant.

[kevinburke]

Apologies, but there’s a zero percent chance it’s my error. If it was, my test suite would fail and go doc would fail to render the example. Neither of those things happen.

I’ll find the root cause and submit a patch today to make that clear.

[kirillDanshin]

basically, no, in your case it will work with tests and go doc, but go vet will indicate it as an error and will be right

[kevinburke]

It's this issue: https://github.com/gojp/goreportcard/issues/187

Also reported here: https://github.com/gojp/goreportcard/issues/113

and here: https://goreportcard.com/report/github.com/arsham/expvastic

also here: https://github.com/gojp/goreportcard/issues/113#issuecomment-331786252

here's a discussion of an error in the handling of this situation inside "go vet": https://github.com/golang/go/issues/22530. The result of the discussion is you should be using "go vet" directly, not "go tool vet". Which is another error, since goreportcard indicates it's using "go vet". Furthermore, I can't reproduce the error locally using "go tool vet" either, though this may be because I'm running tip.

So, again, it's a problem with the tool, not a problem in my code. I only wish I'd bet you money on that outcome first.

[kevinburke]

I'm not 100% sure what this is referring to:

golint edge cases can be fixed, but I even didn't seen before real-life use of that like piece of sh code, that can trigger that bugs.

I ran into one of the "piece of shit" edge cases while trying to update the x/crypto/ssh package: https://go-review.googlesource.com/c/crypto/+/80145#message-1d05ae3ca1b90b6041ab9d5e6edfc88e6e89f38c

Others have as well, multiple times:

• https://github.com/golang/lint/issues/354
• https://github.com/golang/lint/issues/290
• https://github.com/golang/lint/issues/202
• https://github.com/golang/lint/pull/181

I'd also note the Code of Conduct expects you to exercise collaboration and respect in your speech, and marks 'demeaning' speech, such as (incorrectly) labeling code a "piece of shit", as unacceptable. https://github.com/avelino/awesome-go/blob/master/CODE_OF_CONDUCT.md

[kirillDanshin]

note, that you just guessed the continuation of word from a list of options. also, this part of message was a joke and was not meant to be insulting to any Gopher at all.

anyway, it's not changing the subject of this issue

[kevinburke]

Oh, for sure man. There's exactly one word that typically follows "piece of sh", and especially only one word you'd feel the need to cross out and we both know what it is.

Your behavior in this entire thread is embarrassing.

[kevinburke]

Would also love to point you at the "Purpose" disclaimer for golint:

The suggestions made by golint are exactly that: suggestions. Golint is not perfect, and has both false positives and false negatives. Do not treat its output as a gold standard. We will not be adding pragmas or other knobs to suppress specific warnings, so do not expect or require code to be completely "lint-free". In short, this tool is not, and will never be, trustworthy enough for its suggestions to be enforced automatically

Which doesn't seem to match up with my interactions with contributors to this repository.

[kirillDanshin]

@kevinburke I need to also point out that when I use golint/govet output to analyze repository, I assume that they can determine only most generic errors and I checking out every error if I can.
so, basically, we can ignore golint/govet results if we see that there are nothing developer can fix and improve overall code quality.
golint/govet will never replace human analyzing, code review of each file in the repo etc.
it just a tool we use and love for simplifying issue and possible bugs search