Hi Dan,

As a follow-up to this pre-forum-ML-thread, it seems from the API
that only full values can be fetched from LSM. Any chance partial read
similar to the Blob-IO API of SQLite, would be available?

If it was available, would LSM need to chase overflow pages like
SQLite does? I.e. does the way LSM physically store values allow it
to directly jump over to the right page to start reading a far-offset
of a large (multi-MB) value?

I don't even know whether LSM is paged or not I realize.
Is there doc equivalent to SQLite's file-format for LSM.

Thanks, --DD