GeezerVison

common.mustCall variant can be used here.

We could, but I'm not sure we should. This test isn't checking to see if res.end() fires its callback or not. If the response isn't sent, the test will fail. We don't need the extra check. Adding checks that are unrelated to the test case makes it harder for someone reading the code to understand what the test is actually supposed to be testing.

This makes it sound like I'm much more opposed to the idea than I actually am. I'm on the fence, around a -0 or so.

I was a -1 about 10 minutes ago, so 10 minutes from now, I'll probably be an enthusiastic +1.

This test isn't checking to see if res.end() fires its callback or not

correct me if I am wrong, but isn't that the case with most of 100+ http test cases?

• intent of the test does not cover explicit checking on res.end
• but as a matter of practice, they do that check
  if you can generalize a statement around whether http tests should check life cycle events are met or not I can also work towards implementing this across other tests too.

• intent of the test does not cover explicit checking on res.end
• but as a matter of practice, they do that check

I'm not seeing that with res.end() calls. I'm using grep and doing a naive/superficial check, so results may be imperfect but close enough to draw reasonable conclusions:

• Almost none of the 200+ calls to res.end() in parallel/test-http* have a callback wrapped in mustCall(). The majority of them are like the call here where res.endI() is called and no callback is provided.
• Of those that do wrap a callback in muscCall(), only one file checks empty callback: test-http-outgoing-end-multiple.js. That makes sense for that test because it is literally checking that calling res.end() multiple times will work as expected.

I don't think we should add a check for a callback here that we're not using. Again, I'm not strongly opposed, but I am mildly resistant. 😀

I used this command grep -e "res.on('end" -e "response.on('end" test-http*.js | grep "mustCall" | wc -l and got around 50 instances. Unfortunately I don't have the insight to validate their context. so I will leave it at that. Thank you!