fix: handle env-vars case sensitive

[KnisterPeter]

Closes #1488

[github-actions]

🦙 MegaLinter status: ✅ SUCCESS

Descriptor	Linter	Files	Fixed	Errors	Elapsed time
✅ EDITORCONFIG	editorconfig-checker	2		0	0.01s
✅ REPOSITORY	gitleaks	yes		no	2.49s
✅ REPOSITORY	git_diff	yes		no	0.0s
✅ REPOSITORY	secretlint	yes		no	0.9s

See errors details in artifact MegaLinter reports on CI Job page
Set VALIDATE_ALL_CODEBASE: true in mega-linter.yml to validate all sources, not only the diff

MegaLinter is graciously provided by

[codecov]

Codecov Report

Merging #1493 (c29694a) into master (4f8da0a) will increase coverage by 3.68%.
The diff coverage is 67.46%.

@@            Coverage Diff             @@
##           master    #1493      +/-   ##
==========================================
+ Coverage   57.50%   61.19%   +3.68%     
==========================================
  Files          32       45      +13     
  Lines        4594     6957    +2363     
==========================================
+ Hits         2642     4257    +1615     
- Misses       1729     2400     +671     
- Partials      223      300      +77     

Impacted Files	Coverage Δ
pkg/common/file.go	0.00% <0.00%> (ø)
pkg/container/docker_logger.go	52.08% <ø> (ø)
pkg/container/host_environment.go	0.00% <0.00%> (ø)
pkg/container/parse_env_file.go	0.00% <0.00%> (ø)
pkg/container/util.go	0.00% <0.00%> (ø)
pkg/model/action.go	0.00% <0.00%> (ø)
pkg/model/step_result.go	0.00% <ø> (ø)
pkg/container/docker_run.go	13.58% <11.48%> (+8.04%)	⬆️
...ontainer/linux_container_environment_extensions.go	24.32% <24.32%> (ø)
pkg/model/workflow.go	46.43% <25.25%> (-4.48%)	⬇️
... and 38 more

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[jsoref]

Technically they're case-preserving on Windows and case-sensitive everywhere else.

[ChristopherHX]

Yes, but it requires a mass change to replace the golang map object with a custom interface.

[ChristopherHX]

This basically just reverts a previous change and adds a test.

[jsoref]

Right. As long as we're generally evolving to match GHA and people understand that what is being done here isn't the final state.

[KnisterPeter]

@jsoref I do understand the difference but what do you see as limitation with the current implementation? We are probably not converting case of env-var somewhere.
The only issue I see would be that a user sets an env-var with a different case then an existing one and expects it to be overwritten. Is this what you are aiming at?

Feel free to open PRs to improve on this topic.

[jsoref]

On Windows, yes, that.

Note: I don't really use Windows for my GHA stuff. I just dabble in it around the edges and know most of the quirks (and have filed bugs about the GHA docs which were wrong on these same points).

I'm just slightly worried that people will assume that the tests and commit descriptions define normative behavior for all platforms where as they only define the closest-desired behavior for Linux (and macOS).

[catthehacker]

I'm just slightly worried that people will assume that the tests and commit descriptions define normative behavior for all platforms where as they only define the closest-desired behavior for Linux (and macOS).

They shouldn't trust any software then...