| From: | Andres Freund <andres(at)anarazel(dot)de> |
|---|---|
| To: | Ashwin Agrawal <aagrawal(at)pivotal(dot)io> |
| Cc: | Robert Haas <robertmhaas(at)gmail(dot)com>, PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org> |
| Subject: | Re: heapam_index_build_range_scan's anyvisible |
| Date: | 2019-06-10 21:56:38 |
| Message-ID: | 20190610215638.4g2a5xpss33iufke@alap3.anarazel.de |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
Hi,
On 2019-06-10 13:48:54 -0700, Ashwin Agrawal wrote:
> While at it might be helpful and better to also decouple HeapTuple
> dependency for IndexBuildCallback.
Indeed.
> Currently, all AM needs to build HeapTuple in
> index_build_range_scan function. I looked into all the callback functions
> and only htup->t_self is used from heaptuple in all the functions (unless I
> missed something). So, if seems fine will be happy to write patch to make
> that argument ItemPointer instead of HeapTuple?
I wonder if it should better be the slot? It's not inconceivable that
some AMs could benefit from that. Although it'd add some complication
to the heap HeapTupleIsHeapOnly case.
Greetings,
Andres Freund
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Thomas Munro | 2019-06-10 22:00:47 | Re: POC: Cleaning up orphaned files using undo logs |
| Previous Message | David Rowley | 2019-06-10 21:45:16 | Re: Should we warn against using too many partitions? |