Traceback (most recent call last): File "python/mozbuild/mozbuild/codecoverage/lcov_rewriter.py", line 731, in <module> main() File "python/mozbuild/mozbuild/codecoverage/lcov_rewriter.py", line 728, in main rewriter.rewrite_file(f, args.output_suffix) File "python/mozbuild/mozbuild/codecoverage/lcov_rewriter.py", line 675, in rewrite_file assert os.path.isfile(source_file), "Couldn\'t find mapped source file at %s!" % source_file AssertionError: Couldn\'t find mapped source file at /home/marco/Documenti/FD/mozilla-unified/browser/extensions/onboarding/content/modules/OnboardingTourType.jsm! The SF entry is "resource://onboarding/modules/OnboardingTourType.jsm". The module can be found at "browser/extensions/onboarding/OnboardingTourType.jsm".

This is causing the rewriting to fail, as this SF is in a lot of lcov files. Chris, can you take a look?

This is an example lcov file containing OnboardingTourType.jsm.

In the chrome-map.json file we have these entries: "resource://onboarding/": [ "dist/bin/browser/features/onboarding@mozilla.org/chrome/content" and: "dist/bin/browser/features/onboarding@mozilla.org/chrome/content/": [ "/home/marco/Documenti/FD/mozilla-unified/browser/extensions/onboarding/content/*",

There's also: "dist/bin/browser/features/onboarding@mozilla.org/chrome/content/modules/": [ "/home/marco/Documenti/FD/mozilla-unified/browser/extensions/onboarding/*.jsm", I suppose the rewriter should have used this entry instead of the one from comment 3.

Using the longest prefix match in _find_install_prefix fixes the issue. Do you believe it could cause other issues to crop up?

Comment on attachment 8885281 [details] [diff] [review] Patch Review of attachment 8885281 [details] [diff] [review]: ----------------------------------------------------------------- This makes sense and looks like it will work for now. I guess this could still be an issue if we end up with a jar.mn with something like: content/modules/ (*.jsm) content/modules/ (foo/*.jsm) in this case I think we'd only be able to tell which line was responsible for installing the file by trying each pattern, but I'm not sure that's a concern in practice.

Pushed by mcastelluccio@mozilla.com: https://hg.mozilla.org/integration/mozilla-inbound/rev/3b5eec74e7e5 Use longest prefix match when rewriting lcov entries. r=chmanchester

This is happening again, I need to figure out why.

Apparently it's happened again because a task was triggered for an old revision of mozilla-central. So it's actually fine.