typecheck: possible wrong cut

[ldez]

• Yes, I'm using a binary release within 2 latest major releases. Only such installations are supported.
• Yes, I've searched similar issues on GitHub and didn't find any.
• Yes, I've included all information below (version, config, etc).
• Yes, I've tried with the standalone linter if available. (https://golangci-lint.run/usage/linters/)

#1861 (comment)

Description of the problem

Your description

Version of golangci-lint

$ golangci-lint --version
golangci-lint has version 1.39.0 built from 9aea4aee on 2021-03-26T08:02:53Z

Config file

$ cat .golangci.yml
# paste output here

Go environment

$ go version && go env
# paste output here

Verbose output of running

$ golangci-lint cache clean
$ golangci-lint run -v
# paste output here

Code example or link to a public repository

// add your code here

[ldez]

@thebeline could you provide the information to fill the issue?

• Config file
• Verbose output of running
• Go environment
• example or a public repository

[ldez]

@thebeline I think that your problem is related to something else:

• are you using go1.16?
• are you using go modules?

[ldez]

I think it's about https://github.com/thebeline/OctoScreen/tree/IMP/GolangLinter or https://github.com/thebeline/OctoScreen/tree/MESS/GolangLinter

@thebeline could you confirm?

[thebeline]

@ldez - Certainly. I need to draft up an MVP for the issue. But below I will try to describe the steps as I can remember.

• Hardware: Raspberry Pi 3 B
• OS: OctoPi (Debian BUSTER)
• Go Package Checkout
  • Clone https://github.com/thebeline/OctoScreen.git into /home/pi/OctoPrint
  • Checkout IMP/GolangLinter
• Golang Setup
  • Raspbian (or, at least OctoPi) ships with an old Go (1.11, I believe)
  • We want Golang 1.15.10, as 1.16.0 behaves oddly with the version of gotk3 we are using.
  • Remove Golang if installed.
  • Install Golang 1.15.10
    • $ wget https://golang.org/dl/go1.15.10.linux-armv6l.tar.gz
    • $ sudo tar -C /usr/local -xzf go1.15.10.linux-armv6l.tar.gz
  • Install GolangCI-Lint
    • curl -sSfL https://raw.githubusercontent.com/golangci/golangci-lint/master/install.sh | sh -s -- -b $(go env GOPATH)/bin v1.39.0
  • Install the build dependencies (not all are needed, but this was in my history, I think only GTK and Cairo are needed)
    • sudo apt-get update && sudo apt-get install -y --no-install-recommends build-essential git-buildpackage fakeroot libcairo2-dev libgtk-3-dev git devscripts
  • Preconfigure some build flags (GTK 2.24 is bugged as of January, flag for 2.22)
    • $ go env -w GOFLAGS=-tags=gtk_3_22,pango_1_42,gdk_pixbuf_2_38,glib_2_58
• Run golangci-lint
  • $ /home/pi/go/bin/golangci-lint run --new=false --issues-exit-code 0 -v --modules-download-mode vendor

There are probably some red flags in there, but note that it does run in GitHubActions.

Verbose Output HEAD (errors as described follow):

/home/pi/go/bin/golangci-lint run --new=false --new-from-rev= --issues-exit-code 0 -vvv --modules-download-mode vendor
INFO [config_reader] Config search paths: [./ /home/pi/OctoScreen /home/pi /home /]
INFO [config_reader] Used config file .golangci.yml
INFO [lintersdb] Active 9 linters: [dogsled errorlint gofumpt makezero nolintlint unparam unused whitespace wsl]
INFO [loader] Go packages loading at mode 575 (imports|types_sizes|compiled_files|deps|files|exports_file|name) took 1.942055235s
INFO [runner/filename_unadjuster] Pre-built 0 adjustments in 56.589407ms
INFO [linters context/goanalysis] analyzers took 4.629646ms with top 10 stages: errorlint: 2.578153ms, buildir: 364.737µs, whitespace: 327.289µs, makezero: 219.216µs, gofumpt: 205.726µs, the_only_name: 185.988µs, dogsled: 156.249µs, nolintlint: 143.54µs, buildssa: 136.404µs, isgenerated: 95.729µs
INFO [runner] Processors filtering stat (out/in): max_per_file_from_linter: 2220/2220, path_prefixer: 2220/2220, nolint: 2220/2220, path_prettifier: 2220/2220, exclude: 2220/2220, exclude-rules: 2220/2220, diff: 2220/2220, max_same_issues: 2220/2220, source_code: 2220/2220, path_shortener: 2220/2220, filename_unadjuster: 2220/2220, severity-rules: 2220/2220, identifier_marker: 2220/2220, uniq_by_line: 2220/2220, max_from_linter: 2220/2220, skip_files: 2220/2220, skip_dirs: 2220/2220, autogenerated_exclude: 2220/2220, sort_results: 2220/2220, cgo: 2220/2220
INFO [runner] processing took 580.551429ms with stages: identifier_marker: 275.213929ms, exclude-rules: 224.839465ms, nolint: 45.997165ms, source_code: 11.651398ms, path_prettifier: 9.721106ms, sort_results: 5.710415ms, path_shortener: 2.261071ms, filename_unadjuster: 1.245666ms, cgo: 1.150563ms, severity-rules: 1.018845ms, max_per_file_from_linter: 932.7µs, autogenerated_exclude: 788.691µs, max_same_issues: 5.625µs, uniq_by_line: 2.448µs, exclude: 2.448µs, skip_files: 2.135µs, path_prefixer: 2.083µs, skip_dirs: 2.083µs, max_from_linter: 1.876µs, diff: 1.717µs
INFO [runner] linters took 11.045904951s with stages: goanalysis_metalinter: 10.464684572s

After alll of this, I think I should clarify: Locally, I know this error is caused because gccgo is not installed.

That is my mistake.

What I am actually raising as an issue:

• Like I said in my previous comment, I think converting this condition to a "linter error" hides underlying issues (that I would have seen much more clearly in 1.38)
• This triggers LINTER_COUNT+1 times.
  • If it is known that this error state will prevent a linter from continuing, maybe don't continue (although I know that is probably difficult to predict)
  • "Unique by line and character" seems like it would be helpful, just in general.
• By and large, it would seem that if golangci-lint were able to detect catastrophic environment issues (such as cgo preprocessing failed, or failing to load basic go standard libraries), there have to be more helpful debug messages.

MESS/GolangLinter can be ignored for now, but yes, IMP/GolangLinter was where I was working.

[thebeline]

With regards to the IMP/GolangLinter branch, please note that the majority of those commits were just seeing how things behaved.

One thing that I was looking at was related to: https://github.com/golangci/golangci-lint-action/issues/192

Which, as I go down the rabbit hole, I now realize has nothing to do with the action, but moreso with linter it's self.

Actually, I have never run the linter locally, which is why I hadn't seen this issue until now. I only just started trying to run it locally because I was going to try poking things to figure out how the linter worked, and why exactly it was behaving as I had mentioned in the Action Issue I opened.

[ldez]

Your problem in golangci/golangci-lint-action#192 is related to #1490, it's not related to the github action

[thebeline]

Ah, it would seem, more specifically: #1490 (comment)

A function like mergeLineIssues seems like it might have something to do with things I mentioned. :-)

Still, it seems odd that if golangci-lint, under the hood, knows that the env is broken, why does it keep going?

[ldez]

golangci-lint doesn't know if the env is broken, golangci-lint just run Go analysis and run the linters.
It just displays the report from Go analysis: each linter has differents requirements and the fact to compile is not required by all the linters.