Hi Giorgio Zoppi, It seems that only Pageheader has its serialize and deSerialize method in its class. Chunk and Page's serialize method is encapsulated in ChunkWriter and PageWriter.
IChunkLoader in ChunkMetadata is used in iotdb-server to load the corresponding chunk from disk or memory, depending on whether it is DiskChunkLoader or MemChunkLoader. In our query processing, we first get TimeseriesMetadata and then if it is satisfied with the filter, we use IChunkMetadataLoader in TimeseriesMetadata to get a list of ChunkMetadata. Then we iterate each ChunkMetadata, and if it is satisfied with the filter, we use IChunkLoader in ChunkMetadata to load chunk which contains the final page data. Best, ----------------------------- Yuan Tian On Tue, Oct 12, 2021 at 6:43 PM Giorgio Zoppi <giorgio.zo...@gmail.com> wrote: > Hello IOTDBers, > I am going a bit further. Basically in your model each element(chunk, page. > pageheader) serializes itself. This violates the separation of concern > between serialization and data (usually serializers use > annotation/metaprogramming or introspection), but we're fine. My question > is about the chunk loaders. In each metaindex node you have chunk loaders, > why are they there? Which is their real use? (Looks to me a chain of > responsibility pattern) Are there to load when a chunk of data/metadata? > Thank you, > Best Regards, > Giorgio. > > > > -- > Life is a chess game - Anonymous. >
