Here's an example of the error that I get with Azul that I don't get with non-M1 JDKs: - convert Base64 JSON back to rpo Avro *** FAILED *** I think that is because there are some native things written in C++ for Parquet/Avro, and they are not compiled for M1.įor this specific reason I am forced to use the non-M1 JDK, which is slow. For example, when the test writes to a Parquet/ Avro file, it fails. But I noticed that certain tests started failing after the switch. And I started using the Azul JDK for my development workflow. Especially when some Java code needs to call C++ code.įor example, I am a big data developer. I just wanted to say that, while the Azul JDK works natively on Apple M1 and its speed is great, there are still issues. | | 18.ea.3 | open | | 18.ea.3-openĬhoose one and install it using the command sdk install java IDENTIFIER, i.e.: Vendor | Use | Version | Dist | Status | IdentifierĪzul Zulu | | 16.0.1 | zulu | | 16.0.1-zuluīellSoft | | 16.0.1 | librca | | 16.0.1-librca Set sdkman_rosetta2_compatible=false (see sdkman config)Īfter that, you will see a list of compatible with M1 JDKs: sdk list java You can install the Java JDK using sdkman (see sdkman install): vim. Note that Microsoft's been working on the OpenJDK branch of AArch64 (for ARM-based Windows 10) for a while, which goes back to: A lot of the hard work was already done.
0 Comments
Leave a Reply. |