cc @sipsorcery

@ken2812221 isn't NDEBUG desirable for msvc release builds? Without it assert statements will terminate the programs.

Update: After looking over #15391 I now see the origin of the removal. I guess it's a design decision for the bitcoin-core programs but on Windows it's a real pain when bitcoin-qt crashes as the result of an assertion. It's very rare to see that type of behaviour on Windows apps. The preferred approach is to throw an exception and at least display a usable error message to users before terminating.

You can't build it after #15391 merged. The CI pass because I disabled them by the appveyor.yml

@sipsorcery If an assertion is false, aborting is often far better than the alternative (with assumptions violated, we may accept an invalid chain for example...).

How could this compile locally, but then suddenly stop to compile with #15391, given that we have the same compile time check in validation.cpp?

Maybe @sipsorcery just delete them in libbitcoin_server.vcxproj in his first PR?

tACK 8a1f0a3 3ec56be.

@MarcoFalke if your comment was for me then I didn't have any problems compiling with msvc before or after this PR. Instead I was just having a little moan about the unfriendly way asserts terminate bitcoin-qt. But that decision was made by a wiser person(s) than me so I'll track down the root cause instead.