Currently InputStream.skip(...) functionality uses same code path as reads, which works functionally, but does unnecessary work of decoding. For some use cases it would be really great to do "coarse skip", since format allows one to completely skip a chunk without decoding, as the length is known on per-chunk basis. Depending on underlying storage format, this could possibly allow skipping some reads too.
Once implemented, it would be also be good to see if a file-based approach could do even better and use random-access for access. But first steps first.
Currently InputStream.skip(...) functionality uses same code path as reads, which works functionally, but does unnecessary work of decoding. For some use cases it would be really great to do "coarse skip", since format allows one to completely skip a chunk without decoding, as the length is known on per-chunk basis. Depending on underlying storage format, this could possibly allow skipping some reads too.
Once implemented, it would be also be good to see if a file-based approach could do even better and use random-access for access. But first steps first.