Boeing didn't conduct end-to-end tests on Starliner before its failed flight

I don’t actually have any fully pure unit tests in my codebase at the moment. The exception is some algorithmic stuff which has a bunch of tests. In practice, I’ve almost never caught a bug because of a unit test except in an algorithm - but integration tests involving 3-4 small components are great.

<i>I really don&#x27;t want anyone to have the impression that this team tried to take shortcuts</i><p>- man who let his team take shortcuts<p><i>NASA is still thinking whether to allow Boeing to proceed with its first manned flight to prevent delays or to require the company to successfully complete an unmanned flight first.</i><p>Yeah, it&#x27;s probably fine now. I mean, ugh, look at all these passing tests!<p>It&#x27;s staggering to me that <i>Boeing</i> is confident enough to not want another unmanned fight. After all this, you really want to roll the dice again?

How does one do an end-to-end test on a system that ends up in space?

Unit tests do not negate the need for Integration Tests and they do not negate the need for manual tests. You can get overlapping coverage and slack in certain areas in some cases... medical equipment and aerospace are not areas you can come up short with some of the longer testing cycles.