Delete key fails when you use set a subkey for the fingerprint input

[josecelano]

Behaviour

If you use the input fingerprint with the fingerprint of the subkey, the Post import GPG key step tried to delete the key using the fingerprint of the subkey. I suppose you have to use the KeyID (fingerprint of the primary key).

Steps to reproduce this issue

They were described in this PR.

Expected behaviour

I should delete the key.

Actual behaviour

It does not delete the key and it was correctly imported into the keyring.

Configuration

• Repository URL (if public): https://github.com/Nautilus-Cyberneering/chinese-ideographs-website
• Build error URL (if public): https://github.com/Nautilus-Cyberneering/chinese-ideographs-website/runs/5336399157?check_suite_focus=true#step:21:3

Logs

[josecelano]

I'm getting this error now:

It seems we have to remove the --batch option here.

Lint to the error: https://github.com/Nautilus-Cyberneering/nautilus-librarian/runs/5394342708?check_suite_focus=true#step:25:3

ghaction-import-gpg/src/gpg.ts

Line 211 in 87adbd8

.getExecOutput('gpg', ['--batch', '--yes', '--delete-secret-keys', fingerprint], {

The test did not catch this case because it's using the fingerprint. We can reopen the issue, add a new test for that case and fix it.

ghaction-import-gpg/__tests__/gpg.test.ts

Line 133 in 87adbd8

describe('deleteKey', () => {

[t3chguy]

I am experiencing this in v5 of the action when passing & loading only a signing subkey. It tries to delete using the fingerprint of the pubkey which is discovered via the subkey.

t3chguy@Michael-t3chguy-MBP ~> gpg --list-keys --with-subkey-fingerprints D7B0B66941D01538
pub   rsa4096 2019-04-15 [SC] [expires: 2024-04-13]
      12D4CD600C2240A9F4A82071D7B0B66941D01538
uid           [ unknown] riot.im packages <packages@riot.im>
sub   rsa3072 2019-04-15 [S] [expires: 2023-04-15]
      75741890063E5E9A46135D01C2850B265AC085BD

[amorey]

I'm seeing the same error using the latest version (v6.3.0). Is it possible that the bugfix was reverted? The PR that fixed the issue modified the code to use stateHelper.keyID for deletion but now it's using stateHelper.fingerprint again:
https://github.com/crazy-max/ghaction-import-gpg/blob/master/src/main.ts#L141