assertions | Fluent assertion-style functions | Assertion library

I think I worked out the root cause of the issues I was experiencing, I'm posting the solution in case someone else runs into similar issues. I had the following tests configuration:

Neither assertion can ever fail, thanks to ISO C++'s "release sequence" rules. This is the formalism that provides the guarantee you assumed must exist in your last paragraph.

The only stores to val are release-stores with the appropriate bits set, done after the corresponding store to f1 or f2. So if thread_3 sees a value with 1 bit set, it has definitely synchronized-with the writer that set the corresponding variable.

And crucially, they're each part of an RMW, and thus form a release-sequence that lets the acquire load in thread_3 synchronize-with both CAS writes, if it happens to see val == 3.

(Even a relaxed RMW can be part of a release-sequence, although in that case there wouldn't be a happens-before guarantee for stuff before the relaxed RMW, only for other release operations by this or other threads on this atomic variable. If thread_2 had used mo_relaxed, the assert on f2 could fail, but it still couldn't break things so the assert on f1 could ever fail. See also What does "release sequence" mean? and https://en.cppreference.com/w/cpp/atomic/memory_order)

If it helps, I think those CAS loops are fully equivalent to val.fetch_or(1, release). Definitely that's how a compiler would implement fetch_or on a machine with CAS but not an atomic OR primitive. IIRC, in the ISO C++ model, CAS failure is only a load, not an RMW. Not that it matters; a relaxed no-op RMW would still propagate a release-sequence.

(Fun fact: x86 asm lock cmpxchg is always a real RMW, even on failure, at least on paper. But it's also a full barrier, so basically irrelevant to any reasoning about weakly-ordered RMWs.)

After some more search arounds, found this is a known issue in react-native-reanimated. As their website points out

Please note that Reanimated 2 doesn't support remote debugging, only Flipper can be used for debugging.

Another github issue also pointed out this issue

This is expected, you can't use remote debugging with turbomodules (which Reanimated v2 is using). Check out Flipper to debug your app.

https://docs.swmansion.com/react-native-reanimated/docs/#known-problems-and-limitations

https://github.com/software-mansion/react-native-reanimated/issues/1990

Removing this library fixed the issue.

1. Remove the react-native-reanimated dependency in package.json
2. Remove related code in android's MainApplication.java
3. yarn install or npm install
4. Go to the ios folder and run pod install
5. Go the the android folder and run ./gradlew clean
6. Rebuild the app. yarn android and yarn ios

Another alternative is to use Flipper for debugging instead.

I've just had this exact issue with a .NET Framework 4.8 console app. Would build fine locally but failed the build step in the Azure DevOps pipeline.

Turns out that pipeline was using the vs2017-win2016 vm. Bumping it up to windows-2019 - which used Visual Studio 2019/later version of MSBuild - sorted the issue.

If you want to test that a "thing" is deleted, make a fixture without teardown, delete it in the test, then assert if it is deleted.

The entire Debug.Assert line is optimised away when you build in Release mode. So:

As you already figured out in the comments, the tricky part about FTP passive mode is that the server uses another port (not 21) for communication. In the docker image you're using, it's a port from the 21000-21010 range by default. So you need to publish (expose) these additional container ports. In docker run command you used -p 21000-21010:21000-21010 for that.

However, Testcontainers library is designed to publish to random host ports to avoid the problem, when a desired fixed port (or a range of ports) is already occupied on the host side. In case of FTP passive mode random ports on the host side cause problems, because afaik you can't instruct the ftp client to override the port, which FTP server returned for the passive mode. You'd need something like ftpClient.connect("localhost", ftp.getMappedPort(PORT)); but for passive mode ports as well.

Therefore the only solution I see here is to use a FixedHostPortContainer. Even though it's marked as deprecated and not recommended to use because of the mentioned issues with occupied ports, I think this is a valid use case for it here. FixedHostPortGenericContainer allows to publish fixed ports on the host side. Something like:

I got the same problem. Open your terminal and write this:

Old x86-64 CPUs don't support lzcnt, so rustc/llvm won't emit it by default. (They would execute it as bsr but the behavior is not identical.)

Use -C target-feature=+lzcnt to enable it. Try.

More generally, you may wish to use -C target-cpu=XXX to enable all the features of a specific CPU model. Use rustc --print target-cpus for a list.

In particular, -C target-cpu=native will generate code for the CPU that rustc itself is running on, e.g. if you will run the code on the same machine where you are compiling it.