Обсуждение: Re: [WIP]Vertical Clustered Index (columnar store extension) - take2
Hi Peter, I've noticed there are changes in Postgres code v4 patch that rollback the commit [1]. That commit optimizes TupleHashEntryData struct size and amount of memory allocations which improves performance (see discussion [2]). Can we use leave TupleHashEntryData as is and make new VCI-specific struct that contains TupleHashEntryData member and an additional pointer or make VCI use TupleHashEntryGetAdditional()? [1] https://www.postgresql.org/message-id/attachment/174581/v9-0004-Remove-additional-pointer-from-TupleHashEntryData.patch [2] https://www.postgresql.org/message-id/flat/817d244237878cebdff0bc363718feaf49a1ea7d.camel%40j-davis.com ------- Regards, Timur Magomedov
Hi Timur. On Thu, May 29, 2025 at 11:30 PM Timur Magomedov <t.magomedov@postgrespro.ru> wrote: > > Hi Peter, > I've noticed there are changes in Postgres code v4 patch that rollback > the commit [1]. That commit optimizes TupleHashEntryData struct size > and amount of memory allocations which improves performance (see > discussion [2]). > Can we use leave TupleHashEntryData as is and make new VCI-specific > struct that contains TupleHashEntryData member and an additional > pointer or make VCI use TupleHashEntryGetAdditional()? > > [1] > https://www.postgresql.org/message-id/attachment/174581/v9-0004-Remove-additional-pointer-from-TupleHashEntryData.patch > [2] > https://www.postgresql.org/message-id/flat/817d244237878cebdff0bc363718feaf49a1ea7d.camel%40j-davis.com Thank you for noticing and reporting this! It was not intentional to roll back changes to core PostgreSQL. These VCI patches originated from an older forked source, so it seems this reversion was inadvertently introduced during the rebasing process. We’ll aim to correct this in a future patch. ====== Kind Regards, Peter Smith. Fujitsu Australia
On Thu, May 29, 2025 at 11:30 PM Timur Magomedov <t.magomedov@postgrespro.ru> wrote: > > Hi Peter, > I've noticed there are changes in Postgres code v4 patch that rollback > the commit [1]. That commit optimizes TupleHashEntryData struct size > and amount of memory allocations which improves performance (see > discussion [2]). > Can we use leave TupleHashEntryData as is and make new VCI-specific > struct that contains TupleHashEntryData member and an additional > pointer or make VCI use TupleHashEntryGetAdditional()? > Thanks for the report. This is fixed in v6. ====== Kind Regards, Peter Smith. Fujitsu Australia