Closed tigarmo closed 1 year ago
Merging #86 (c65a965) into hotfix/1.1.1 (b1c20e2) will decrease coverage by
0.58%
. The diff coverage is82.85%
.:exclamation: Current head c65a965 differs from pull request most recent head d7fef04. Consider uploading reports for the commit d7fef04 to get more accurate results
@@ Coverage Diff @@
## hotfix/1.1.1 #86 +/- ##
================================================
- Coverage 90.51% 89.94% -0.58%
================================================
Files 13 13
Lines 664 686 +22
Branches 133 140 +7
================================================
+ Hits 601 617 +16
- Misses 47 51 +4
- Partials 16 18 +2
Impacted Files | Coverage Δ | |
---|---|---|
craft_archives/repo/apt_key_manager.py | 94.44% <82.85%> (-3.92%) |
:arrow_down: |
This commit relaxes the restriction of the provided key contents having only a single (primary) key, and makes AptKeyManager better handle the fact that "gpg import" might print out useful diagnostics messages even if the key contents are imported "successfully".
To do this, we capture gpg's output and add a new optional parameter for the desired "target" key fingerprint (the key that we desire to import even if other keys exist in the contents). This fingerprint is used when naming the imported file, and then:
1) If we don't have a desired fingerprint, proceed as before. 2) If we have a desired fingerprint, check for it in the imported file, failing if it's not there.
The motivation for this change is that there are keys "in the wild" that generate gpg error messages but still contain a valid, non-expired desired key. This was supported before but regressed when we moved away from apt-key.
This commit is the second one on the road to fixing the "multiple keys in the same file" regression.
tox
?