yarn:update add a handled error for missing tags #8389
Conversation
|
Any way to capture this in a test? Would that be valuable? |
I'll take a look at existing tests and see what we have for other resolvability errors. |
| @@ -295,7 +295,8 @@ def handle_yarn_lock_updater_error(error, yarn_lock) | |||
| handle_timeout(error_message, yarn_lock) if error_message.match?(TIMEOUT_FETCHING_PACKAGE) | |||
|
|
|||
| if error_message.start_with?("Couldn't find any versions") || | |||
| error_message.include?(": Not found") | |||
| error_message.include?(": Not found") || | |||
There was a problem hiding this comment.
minor: is this issue only specific to yarn or does it impact pnpm and npm too?
There was a problem hiding this comment.
It is a specific yarn error message. There might be specific pnpm and npm error messages that are the "same" but they might already be matched by regex/strings in the respective code.
There was a problem hiding this comment.
The way you explained this, it sounds like maybe Dependabot::GitDependencyReferenceNotFound is a better fit for this error?
Regarding testing this, my initial approach is normally:
- Introduce a dummy change in a existing analogous regex/string.
- Rerun tests and see what breaks.
- Use the failing test as a starting point for a new test.
In the case of git dependencies, if the version is not tagged correctly the error is not caught as a resolvability error. This adds handling for the error message if a release is not tagged as expected.
pavera
force-pushed
the
pavera/yarn_update_missing_tags
branch
from
f2db961
to
ef809af
Compare
|
@pavera Any comments on my previous feedback? |
|
I'm good with merging this without tests if this is hard to test, but it sounds like the user error could be a better one if I understood your explanation correctly? |
In the case of git dependencies, if the version is not tagged correctly the error is not caught as a resolvability error. This adds handling for the error message if a release is not tagged as expected.