Merge ~mertkirpici/charm-juju-lint:lp/1987359 into charm-juju-lint:master-subordinate

PASSED: Continuous integration, rev:6939e63b5c9cae47a91909a689d919050aa30e43
https://jenkins.canonical.com/bootstack/job/lp-charm-juju-lint-ci/37/
Executed test runs:
    SUCCESS: https://jenkins.canonical.com/bootstack/job/lp-charm-test-functest/2161/
    None: https://jenkins.canonical.com/bootstack/job/lp-update-mp/1968/

Click here to trigger a rebuild:
https://jenkins.canonical.com/bootstack/job/lp-charm-juju-lint-ci/37//rebuild

Overall LGTM. I've just add an inline note.

Just one in-line question from my side. Otherwise LGTM

PASSED: Continuous integration, rev:ee4d8c2d347ac6624739e9931d2aa2ebb16baf07
https://jenkins.canonical.com/bootstack/job/lp-charm-juju-lint-ci/38/
Executed test runs:
    SUCCESS: https://jenkins.canonical.com/bootstack/job/lp-charm-test-functest/2165/
    None: https://jenkins.canonical.com/bootstack/job/lp-update-mp/1972/

Click here to trigger a rebuild:
https://jenkins.canonical.com/bootstack/job/lp-charm-juju-lint-ci/38//rebuild

LGTM

PASSED: Continuous integration, rev:91d238f00baa7a31b6380cc5024964b2eba6e5ab
https://jenkins.canonical.com/bootstack/job/lp-charm-juju-lint-ci/39/
Executed test runs:
    SUCCESS: https://jenkins.canonical.com/bootstack/job/lp-charm-test-functest/2169/
    None: https://jenkins.canonical.com/bootstack/job/lp-update-mp/1976/

Click here to trigger a rebuild:
https://jenkins.canonical.com/bootstack/job/lp-charm-juju-lint-ci/39//rebuild

> Just one in-line question from my side. Otherwise LGTM

Updated the implementation after your commend in favor of getting rid of the non-standard proxy.json file as a means to communicate proxy settings between processes.

I would like to explain the current mechanism.

The juju-lint snap is run by the wrapper script /usr/local/bin/auto_lint.py, which is installed via the juju-lint charm. During the lifecycle of the juju-lint charm, there are two different ways/context that this auto_lint.py script is run and we need to adjust the proxy settings for both of them separately:

1 - the script is run directly by the charm code in hook context: during install and config-changed handlers, the charm manually runs the script. see lib_libjuju.JujuLintHelper.run_auto_lint(). we handle the proxy by getting it from the hook environment, and injecting into the subprocess environment variables.

2 - the script is run via crond: the charm installs a cron file at /etc/cron.d/juju-lint, that runs the juju-lint script periodically(see config option "lint-frequency"). it is important to note that this code is _not_ running in any hook context. With the new implementation, the proxy variables are written to the /etc/cron.d/juju-lint file during charm installation/config-changed, which then propagate to auto_lint.py, which further propagates into the juju-lint process.

So currently we are supporting the "standard" way of detecting http{,s} proxies: through environment variables without needing any non-standard way of doing inter process communication, namely a custom file like proxy.json.