Review requested:

• @nodejs/loaders

The commit message guidelines stipulates that the word after the subsystem needs to be an imperative verb, could you amend it please?

@aduh95 the character limit makes this quite a challenge 😅

Here's a suggestion: module: avoid throwing when `findPackageJSON` is called on inexistant

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 88.41%. Comparing base (32ff100) to head (c02aebc).
Report is 355 commits behind head on main.

@@            Coverage Diff             @@
##             main   #55822      +/-   ##
==========================================
+ Coverage   88.39%   88.41%   +0.01%     
==========================================
  Files         654      654              
  Lines      187560   187814     +254     
  Branches    36087    36132      +45     
==========================================
+ Hits       165800   166048     +248     
- Misses      14996    15007      +11     
+ Partials     6764     6759       -5     

... and 49 files with indirect coverage changes

Can you add a test?

Yes, I intend to do, but I have to re-figure out how to re-produce it. I'm 100% certain this fixes the issue and does not cause adverse results, so can we land this without the test-case and add it subsequently so users can get the fix rather than wait?

@aduh95 now that I re-read the updated title, I think it's actually not accurate: the issue does not occur when findPackageJSON(nonexistent), but rather when ESMLoader can't resolve the target, which is subtly different.

so can we land this without the test-case and add it subsequently

Let’s not, without a test we cannot detect regressions.

@AugustinMauroy found the reproduction 🎉 https://github.com/AugustinMauroy/issue-node-findpackagejson

fix LGTM but i agree this should probably have a test case before landing