tape-modern

In TAP, each assertion is counted on its own rather than as being part of a larger test. tape-modern was printing the numbers of the calls though. Previously, assertions were logged with the index of their surrounding call. If you use multiple asserts in a single test, you get output like: This would be interpreted as a failed test by compliant TAP parsers, because the planning line lists different test numbers than the test lines do. With this patch, the output is:

Is the API for tape-modern identical to tape? If not, which exports are missing?

Hello @Rich-Harris! I tried using this package, but because it's missing or , I don't think it's possible to test async stuff like it claims. Here is the situation: 1. Testing a HTTP request library. Creates a dummy server that receives a request from the library, and sends a dummy response. 2. I write assertions in both the dummy server and dummy response. These are asynchronous areas. 3. When running the test, it exits without doing any assertions. I assume because it doesn't know how many assertions to expect ahead of time (purpose of ) and there is no mechanism to tell it when to end (purpose of ) It was easy to swap in regular . But I'm curious, am I using the package wrong or does the package have faulty design?