the comma here was correct (as per C++11) and allows adding more lines without touching this one (smaller diffs, less risks of conflicts)

Are all callers of findCookies() going to remember to do path.isEmpty() ? "/" : path?
Wouldn't it be more robust to do this in findCookies itself?

Or to say this another way -- if other callers of findCookies don't do this, and if we don't want to change them (nor kcookieserver), then this adjustment is wrong. possibly?

I don't think so. If I read the cookie specification correctly, an empty path or a path consisting of a single / should be equivalent. However, while KCookieServer::findCookies actually returns an empty string when the path is empty, QNetworkCookie::normalize converts an empty path to a /.

Ah, so it's a Q-K conversion thing, then it's normal that the other users of KCookieServer don't do this. OK.

But then why is it only done in the unittest and not in the actual code?

I'm also confused because you say "findCookies returns..." but this here is about the argument *passed* to findCookies.

Can id.path be empty?

Because of the call to QNetworkCookie::normalize while setting up the test data (line 254), the cookie's path is never empty. So, the cookie added to KCookieServer can have either a real path or a path which is simply /. This is the reason of the conditional here. However, reading the code again I think it's not the best way to do it. I'll rewrite it to make it better and clearer.

In theory, it can. In practice, it shouldn't. The cookies received by QWebEngineCookieStore don't ever have an empty path. The only way to have an empty path is if some other application adds a cookie to KCookieServer and that cookie is then added to QWebEngineCookieStore from KCookieServer when Konqueror starts.