I'm trying to deserialize a Flux of List<CustomObject>. In this case, the grouping of CustomObjects is necessary: each List is a sequence and my later processing needs to take into account the separate sequences.
I've looked at this question which dealt with an array of objects. In fact, I've been able to use their Mono solution approach to retrieve a Mono<List<List>>. This works, but seems quite inefficient. I'd like to start processing each List, rather than requiring the Mono to complete retrieval of the full bulk set. Our full set could be in the realm of 20K or more lists, some of which we're going to discard, so earlier processing would be desirable.
What happens when I try using Flux instead of Mono: I've tried this using both WebClient and Feign - same effects in each. Stepping through the Jackson code, it never pulls back TokenBuffers when dealing with the Flux world. That seems in line with what's in the Spring documentation for Codecs and Jackson JSON - "when decoding to a multi-value publisher (e.g. Flux), each TokenBuffer is passed to the ObjectMapper as soon as enough bytes are received for a fully formed object"
Thoughts? Even better, working examples?