On Tue, Mar 31, 2020 at 11:48:26PM +0200, Steffen Ullrich wrote: > > You should fix your tests not to trigger an unexpected EOF. You > > probably have code now that ignores the current error, you > > shouldn't ignore that error, it's a real error. > > Fixing the tests to only consider an ideal world of nicely behaving peers is > in my opinion the wrong way to go. Instead I think that it is very useful > to have tests which include bad behaved peers since such peers are > unfortunately a reality. And it would be good if IO::Socket::SSL would > provide a defined and stable behavior in such situations and that the tests > check this. As for how this behavior should exactly look like I'm not fully > sure yet and I have to figure this out before OpenSSL 3.0.0 gets released.
I do agree that testing for errors is important. I assumed that some tests failed because they ignored the error, but I guess they can also fail because they a different error than the one they expected. Kurt