Use App::cpm to install perlcritic faster, and clean .perl-cpm cache

[echoix]

The installation of perlcritic left some cache files in the image, about 31 MB. This PR fixes that.
Also, I tried out other installation methods, and to see if I was able to remove the perl-dev dependency (I couldn't).
App::cpm is a fast CPAN module installer, and they claim it is 3x faster than cpanm. It can use multiple threads to parallelize the work to do. This PR uses that.

I used the tool called dive to look into the images, and hyperfine to benchmark building of the perl_perlcritic image.
Results are :

Command	Mean [s]	Min [s]	Max [s]	Relative	Image size [MB]	Space Saved [MB]
docker buildx build --pull --rm -f "linters/perl_perlcritic/Dockerfile_cpanminus" --progress plain --no-cache -t megalinter-only-perl_perlcritic:test-echoix-cpanminus .	78.554 ± 0.821	78.020	79.499	1.49 ± 0.04	380	-
docker buildx build --pull --rm -f "linters/perl_perlcritic/Dockerfile_cpanminus_clean" --progress plain --no-cache -t megalinter-only-perl_perlcritic:test-echoix-cpanminus-clean .	75.322 ± 2.437	73.781	78.131	1.43 ± 0.06	349	31
docker buildx build --pull --rm -f "linters/perl_perlcritic/Dockerfile_cpm" --progress plain --no-cache -t megalinter-only-perl_perlcritic:test-echoix-cpm .	56.371 ± 2.916	53.776	59.526	1.07 ± 0.06	346	34.2
docker buildx build --pull --rm -f "linters/perl_perlcritic/Dockerfile_cpm_pipe" --progress plain --no-cache -t megalinter-only-perl_perlcritic:test-echoix-cpm-pipe .	52.754 ± 1.246	51.494	53.986	1.00	345	35.4
docker buildx build --pull --rm -f "linters/perl_perlcritic/Dockerfile_cpm_pipe_workers2" --progress plain --no-cache -t megalinter-only-perl_perlcritic:test-echoix-cpm-pipe-workers-2 .	58.401 ± 0.637	57.833	59.090	1.11 ± 0.03	345	35.4

Details of run time

hyperfine --min-runs 3  --export-markdown timings_perlcritic5.md  --export-json timings_perlcritic5.json  \
    'docker buildx build --pull --rm -f "linters/perl_perlcritic/Dockerfile_cpanminus" --progress plain --no-cache -t megalinter-only-perl_perlcritic:test-echoix-cpanminus . ' \
    'docker buildx build --pull --rm -f "linters/perl_perlcritic/Dockerfile_cpanminus_clean" --progress plain --no-cache -t megalinter-only-perl_perlcritic:test-echoix-cpanminus-clean . ' \
    'docker buildx build --pull --rm -f "linters/perl_perlcritic/Dockerfile_cpm" --progress plain --no-cache -t megalinter-only-perl_perlcritic:test-echoix-cpm . '  \
    'docker buildx build --pull --rm -f "linters/perl_perlcritic/Dockerfile_cpm_pipe" --progress plain --no-cache -t megalinter-only-perl_perlcritic:test-echoix-cpm-pipe . ' \
    'docker buildx build --pull --rm -f "linters/perl_perlcritic/Dockerfile_cpm_pipe_workers2" --progress plain --no-cache -t megalinter-only-perl_perlcritic:test-echoix-cpm-pipe-workers-2 . '
Benchmark 1: docker buildx build --pull --rm -f "linters/perl_perlcritic/Dockerfile_cpanminus" --progress plain --no-cache -t megalinter-only-perl_perlcritic:test-echoix-cpanminus . 
  Time (mean ± σ):     78.554 s ±  0.821 s    [User: 0.203 s, System: 0.198 s]
  Range (min … max):   78.020 s … 79.499 s    3 runs
 
Benchmark 2: docker buildx build --pull --rm -f "linters/perl_perlcritic/Dockerfile_cpanminus_clean" --progress plain --no-cache -t megalinter-only-perl_perlcritic:test-echoix-cpanminus-clean . 
  Time (mean ± σ):     75.322 s ±  2.437 s    [User: 0.223 s, System: 0.169 s]
  Range (min … max):   73.781 s … 78.131 s    3 runs
 
  Warning: Statistical outliers were detected. Consider re-running this benchmark on a quiet system without any interferences from other programs. It might help to use the '--warmup' or '--prepare' options.
 
Benchmark 3: docker buildx build --pull --rm -f "linters/perl_perlcritic/Dockerfile_cpm" --progress plain --no-cache -t megalinter-only-perl_perlcritic:test-echoix-cpm . 
  Time (mean ± σ):     56.371 s ±  2.916 s    [User: 0.201 s, System: 0.180 s]
  Range (min … max):   53.776 s … 59.526 s    3 runs
 
Benchmark 4: docker buildx build --pull --rm -f "linters/perl_perlcritic/Dockerfile_cpm_pipe" --progress plain --no-cache -t megalinter-only-perl_perlcritic:test-echoix-cpm-pipe . 
  Time (mean ± σ):     52.754 s ±  1.246 s    [User: 0.174 s, System: 0.187 s]
  Range (min … max):   51.494 s … 53.986 s    3 runs
 
Benchmark 5: docker buildx build --pull --rm -f "linters/perl_perlcritic/Dockerfile_cpm_pipe_workers2" --progress plain --no-cache -t megalinter-only-perl_perlcritic:test-echoix-cpm-pipe-workers-2 . 
  Time (mean ± σ):     58.401 s ±  0.637 s    [User: 0.208 s, System: 0.160 s]
  Range (min … max):   57.833 s … 59.090 s    3 runs
 
Summary
  docker buildx build --pull --rm -f "linters/perl_perlcritic/Dockerfile_cpm_pipe" --progress plain --no-cache -t megalinter-only-perl_perlcritic:test-echoix-cpm-pipe .  ran
    1.07 ± 0.06 times faster than docker buildx build --pull --rm -f "linters/perl_perlcritic/Dockerfile_cpm" --progress plain --no-cache -t megalinter-only-perl_perlcritic:test-echoix-cpm . 
    1.11 ± 0.03 times faster than docker buildx build --pull --rm -f "linters/perl_perlcritic/Dockerfile_cpm_pipe_workers2" --progress plain --no-cache -t megalinter-only-perl_perlcritic:test-echoix-cpm-pipe-workers-2 . 
    1.43 ± 0.06 times faster than docker buildx build --pull --rm -f "linters/perl_perlcritic/Dockerfile_cpanminus_clean" --progress plain --no-cache -t megalinter-only-perl_perlcritic:test-echoix-cpanminus-clean . 
    1.49 ± 0.04 times faster than docker buildx build --pull --rm -f "linters/perl_perlcritic/Dockerfile_cpanminus" --progress plain --no-cache -t megalinter-only-perl_perlcritic:test-echoix-cpanminus . 

Details of disk usage

Using dive with a command similar to:

docker run --rm -it -v /var/run/docker.sock:/var/run/docker.sock wagoodman/dive:latest megalinter-only-perl_perlcritic:test-echoix-cpm-pipe

Note that there is missing the -g global flag in the installation screenshots for cpm, but the sizes are the same, only installed in another folder to be found without changing any paths.

Original installation:

Original installation, cleaning /root/.cpanm folder:

Installing and using App::Cpm, cleaning the /root/.perl-cpm folder:

Using cpm directly from a pipe, without installing it:

Using cpm directly from a pipe, without installing it, and limiting to 2 workers, the closest I can get to what it will be on GitHub Actions runners:

I also tested linting a Perl repo locally with the new installation method, and works as the original

Proposed Changes

1. Use cpm directly from pipe to install Perl Critic
2. Clean perl cache in same step

Readiness Checklist

Author/Contributor

• Add entry to the CHANGELOG listing the change and linking to the corresponding issue (if appropriate)
• If documentation is needed for this change, has that been included in this pull request

Reviewing Maintainer

• Label as breaking if this is a large fundamental change
• Label as either automation, bug, documentation, enhancement, infrastructure, or performance

[echoix]

/build

Command run output
Build command workflow started.
Installing dependencies
Running script ./build.sh

[echoix]

/build

Command run output
Build command workflow started.
Installing dependencies
Running script ./build.sh
Build command workflow completed updating files.

[nvuillam]

Nice, thanks :)