Replace nyc with c8.

[mattcollier]

No description provided.

[JSAssassin]

c8 isn't working properly. It's showing jsdocs as code, giving a higher % of code coverage. I'll be looking into it.

Update: It's not working, tried it on a small project that uses esm and was getting 100% code coverage even when some of the code weren't tested. This maybe related to this issue which is still open - bcoe/c8#34 where others have reported about the same issue. The latest comment on it was on Apr.

[dmitrizagidulin]

Looks good, though yeah, we should investigate how to exclude jsdocs.

[davidlehn]

This should have already been just "lcovonly" since ci only needs the coverage report.

Suggested change

	"coverage-ci": "cross-env NODE_ENV=test c8 --reporter=lcov npm test",
	"coverage-ci": "cross-env NODE_ENV=test c8 --reporter=lcovonly npm test",

[davidlehn]

Is this working?

• Looking at nyc reports from 1.2.0 I see what looks correct, with a few missing coverage spots in meters.js.
• The nyc local output seems correct. The index.html shows the same coverage as above.
• The c8 reports look like they have similar line counts for missing coverage on the index page, but the meters.js page has everything green except for what looks like misplaced missing coverage lines that overlap with docs.
• The c8 local index.html just shows everything as 100% with every meters.js line having run 2x, including empty lines, comments, docs, etc. So zero missing coverage shown. This is wrong.
• I don't speak lcov.info, but it's easy to see the nyc and c8 files are very different.

Am I running it wrong or looking at the wrong output?

[davidlehn]

I tried to add the patch from #6 here. It didn't fix the odd coverage output. I didn't look further. It may be that esm has special nyc support not not c8 support?

[davidlehn]

I think this can be closed due to #6 being merged. Fixing c8 support can be revisited if needed.

[dlongley]

Closing, c8 is now used on the main branch.