The JDK parser caches data when parsing, for example, if a thread with a particular ID exists in two chunks, it only reads the data for the ID once.

This reduces overhead, but it can also hide issues where an ID is not written in the second chunk. This can become visible in JMC, since the JMC parser lacks this optimization.

To find issues where an ID is not written, it would be good to create a test that records data in multiple chunks, but compares the result of parsing chunks individually with the result of parsing them all together. It can also check that constants are not null, which is what happens if a constant is not written in the correct chunk.