golangci-lint - 想要“真正”忽略 go 文件，而不是简单地分析它并抑制警告 - 我当前的忽略方法会占用内存并且速度很慢

Question

使用

golangci-lint

，我知道我可以使用

skip-dirs

或

skip-files

设置，但这些只是在最后停止报告，我认为该工具仍然可以“工作”，引用docs：

...要跳过哪些文件：将对其进行分析，但不会报告其中的问题。 ...

有没有办法，比如使用

//nolint

的变体来防止某些文件首先被

golangci-lint

分析，这样我们就不会在它们上浪费构建时资源 - 内存/CPU ？

我还尝试了顶部的行，如

// Code generated by XXX. DO NOT EDIT.

和

//nolint

停止报告，但我认为

golangci-lint

仍在进行分析 - 即，当存在大的“忽略”文件时，工具在高内存下运行缓慢。来自https://golangci-lint.run/usage/false-positives/#nolint-directive我可以做

> //nolint:unparam    
> package pkg

确实没有报告 lint 错误，但我仍然使用过多的内存 例如

13:14 $ golangci-lint run --timeout 30m -v 
INFO [config_reader] Config search paths: [ REMOVED FOR STACKOVERFLOW POST] 
INFO [config_reader] Used config file .golangci.yml 
INFO [lintersdb] Active 17 linters: [deadcode dogsled errcheck gocyclo golint gosimple govet ineffassign misspell nakedret scopelint staticcheck structcheck typecheck unconvert unused varcheck] 
INFO [lintersdb] Active 17 linters: [deadcode dogsled errcheck gocyclo golint gosimple govet ineffassign misspell nakedret scopelint staticcheck structcheck typecheck unconvert unused varcheck] 
INFO [loader] Go packages loading at mode 575 (deps|imports|name|types_sizes|compiled_files|exports_file|files) took 15.003932542s 
INFO [runner/filename_unadjuster] Pre-built 0 adjustments in 831.549179ms 
INFO [runner/unused/goanalysis] analyzers took 15.395924718s with top 10 stages: U1000: 13.12191427s, buildssa: 2.274010448s 
INFO [runner/goanalysis_metalinter/goanalysis] analyzers took 8m25.344883444s with top 10 stages: golint: 6m29.017594865s, ineffassign: 50.003663795s, vrp: 34.694968113s, misspell: 5.91752885s, buildssa: 4.09136586s, SA2003: 1.003697913s, copylocks: 660.780161ms, SA4000: 601.476823ms, SA4018: 510.871677ms, SA4004: 476.466048ms 
INFO [runner] Issues before processing: 28876, after processing: 0 
INFO [runner] Processors filtering stat (out/in): skip_files: 28876/28876, skip_dirs: 28876/28876, identifier_marker: 53/53, path_prettifier: 28876/28876, autogenerated_exclude: 53/28876, exclude: 0/53, filename_unadjuster: 28876/28876, cgo: 28876/28876 
INFO [runner] processing took 60.293318ms with stages: autogenerated_exclude: 33.592255ms, path_prettifier: 18.17708ms, skip_dirs: 4.421717ms, filename_unadjuster: 1.497968ms, cgo: 1.333012ms, identifier_marker: 1.036647ms, exclude: 220.939µs, nolint: 8.996µs, max_same_issues: 1.448µs, uniq_by_line: 592ns, exclude-rules: 544ns, max_from_linter: 467ns, skip_files: 386ns, diff: 362ns, path_shortener: 331ns, source_code: 309ns, max_per_file_from_linter: 265ns 
INFO [runner] linters took 6m47.422489108s with stages: goanalysis_metalinter: 6m28.298523575s, unused: 19.033707326s 
INFO File cache stats: 2 entries of total size 40.4MiB 
INFO Memory: 4114 samples, avg is 7797.4MB, max is 8629.2MB 
INFO Execution took 7m3.337365705s

当我删除大文件时，golangci-lint 表现良好。

Answer 1

Android 的 syzkaller 使用 专用构建标记来排除

golangci-lint

的文件。使用这种方法可以防止解析文件，并显着减少构建期间的内存消耗。

在需要高效跳过的 Go 文件顶部添加

// !codeanalysis

例如

 // AUTOGENERATED FILE
 
 // +build !codeanalysis

 package foo

在

.golangci.yml

文件中...

run:
  deadline: 8m
  skip-dirs:
    - pkg/kd
  # Autogenerated files take too much time and memory to load,
  # even if we skip them with skip-dirs.
  # So we define this tag and use it in the autogenerated files.
  build-tags:
    - codeanalysis

我已经用我的设置对此进行了简要测试，该工具不再占用大量内存。

以下方法可以“功能性”工作，但性能不佳......

• //nolint

• 跳过目录或跳过文件设置

• // Code generated by XXX. DO NOT EDIT.

  显示文件的行已生成