| From: | Alexander Korotkov <a(dot)korotkov(at)postgrespro(dot)ru> |
|---|---|
| To: | Alvaro Herrera <alvherre(at)2ndquadrant(dot)com> |
| Cc: | Nikita Glukhov <n(dot)gluhov(at)postgrespro(dot)ru>, Thomas Munro <thomas(dot)munro(at)gmail(dot)com>, David Steele <david(at)pgmasters(dot)net>, Anastasia Lubennikova <lubennikovaav(at)gmail(dot)com>, PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org> |
| Subject: | Re: [PATCH] kNN for btree |
| Date: | 2019-09-09 20:28:10 |
| Message-ID: | CAPpHfdud0qwzGESyt_WNbzW3AdtRo6Wi6dYNeyh5QSd0jquHew@mail.gmail.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On Sun, Sep 8, 2019 at 11:35 PM Alexander Korotkov
<a(dot)korotkov(at)postgrespro(dot)ru> wrote:
> I'm going to push 0001 changing "attno >= 1" to assert.
Pushed. Rebased patchset is attached. I propose to limit
consideration during this commitfest to this set of 7 remaining
patches. The rest of patches could be considered later. I made some
minor editing in preparation to commit. But I decided I've couple
more notes to Nikita.
* 0003 extracts part of fields from BTScanOpaqueData struct into new
BTScanStateData struct. However, there is a big comment regarding
BTScanOpaqueData just before definition of BTScanPosItem. It needs to
be revised.
* 0004 adds "knnState" field to BTScanOpaqueData in addition to
"state" field. Wherein "knnState" might unused during knn scan if it
could be done in one direction. This seems counter-intuitive. Could
we rework this to have "forwardState" and "backwardState" fields
instead of "state" and "knnState"?
------
Alexander Korotkov
Postgres Professional: http://www.postgrespro.com
The Russian Postgres Company
| Attachment | Content-Type | Size |
|---|---|---|
| 0002-Enable-ORDER-BY-operator-scans-on-ordered-indexe-v15.patch.gz | application/x-gzip | 1.4 KB |
| 0001-Introduce-ammorderbyopfirstcol-v15.patch.gz | application/x-gzip | 2.6 KB |
| 0003-Extract-structure-BTScanState-v15.patch.gz | application/x-gzip | 10.6 KB |
| 0004-Add-kNN-support-to-btree-v15.patch.gz | application/x-gzip | 16.3 KB |
| 0005-Add-btree-distance-operators-v15.patch.gz | application/x-gzip | 12.0 KB |
| 0006-Remove-distance-operators-from-btree_gist-v15.patch.gz | application/x-gzip | 4.9 KB |
| 0007-Add-regression-tests-for-kNN-btree-v15.patch.gz | application/x-gzip | 4.3 KB |
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Alexander Korotkov | 2019-09-09 21:07:24 | Re: Rethinking opclass member checks and dependency strength |
| Previous Message | Peter Geoghegan | 2019-09-09 19:54:08 | Re: [HACKERS] [WIP] Effective storage of duplicates in B-tree index. |