Hi Sergei: About the `page_range` parameter, I'm trying to understand the rationale behind this addition. As I understand it, records_in_range() is to calculate how many rows match the given condition. If so, the existing parameters (index_name, min_key, max_key) seem sufficient for cardinality estimation.
If my understanding is correct, `page_range` isn't information provided to the storage engine, but rather data that the engine needs to populate and return for the optimizer? For example, when the engine reports that first_page and end_page are the same, the index cost might be calculated as slightly lower? This is quite an interesting feature. For our first phase, I think we could temporarily ignore it without causing significant estimation errors. After the first version, I'd like to create examples where page_range is effective and enhance VIDEX's support for page_range. What do you think? :) Best, Rong _______________________________________________ discuss mailing list -- [email protected] To unsubscribe send an email to [email protected]
