Deprecate pointless cookie expiry facades

[franzliedke]

As outlined in #23 (the problem also came up in #22), some of the "utilities" I introduced in #4 / #10 don't make sense. To expire cookies, the Set-Cookie header needs to have the same parameters that were used when creating them.

Therefore, this PR:

• deprecates the pointless / misleading methods
• changes all tests to use better APIs
• updates the documentation (with a warning about this browser requirement)

Sorry for not thinking this through properly in the original PR - and for taking a while to fix this. 😉
Thanks for the great library! 🙌🏼

Fixes #22, #23.

[franzliedke]

One alternative I see:

FigResponseCookies::expire($response, $setCookie);

Hardly more elegant and slightly harder in terms of BC. Your choice. 😃

[franzliedke]

Tests failures:

Failed to retrieve commit parents. If you use a shallow git checkout, please checkout at least a depth of one.

I can only assume that actions/checkout changed its defaults at some point - see actions/checkout#217.

[simensen]

@franzliedke Would you be able to try playing with fetch-depth or other actions/checkout options to see if we can get it working again? The error from ocular seems a bit weird, though, as it specifically says:

If you use a shallow git checkout, please checkout at least a depth of one.

However, based on the actions/checkout docs, the default fetch-depth is currently 1 already? So, not sure if fetch-depth: 2 is what we need or something else?

Based on this stackoverflow answer it might be that we need to run the following additional command right before the ocular run:

            - run: vendor/bin/phpunit --coverage-text --coverage-clover=coverage.clover
              if: ${{ matrix.coverage != 'none' }}

            # new thing to unshallow our checkout as `ocular` needs parent commits
            - run: git fetch --unshallow

            - run: php vendor/bin/ocular code-coverage:upload --format=php-clover coverage.clover
              if: ${{ matrix.coverage != 'none' }}

[franzliedke]

Sure, I would be happy to. Thanks for the additional research.

I will do it in an additional MR, so that

• I can confirm this is independent
• discussions about the changes here don't block fixing the pipeline

[franzliedke]

I will rebase once #46 has been merged.

[franzliedke]

Rebased, and tests are green. 🙂

[simensen]

@franzliedke Thanks!