Re: [HACKERS] WIP: store additional info in GIN index

2013-03-03 Thread Craig Ringer
On 03/04/2013 01:29 AM, Alexander Korotkov wrote: Given the activity level I would like to bounce this patch, either as "returned with feedback" if you want to take another go at it post-9.3, or as "rejected" if you think the idea won't go anywhere. Please let me know how you think

Re: [HACKERS] WIP: store additional info in GIN index

2013-03-03 Thread Alexander Korotkov
On Sun, Mar 3, 2013 at 6:53 PM, Craig Ringer wrote: > The GIN changes don't seem to have progressed in some time, and some of > the most recent activity > (http://archives.postgresql.org/message-id/50bff89a.7080...@fuzzy.cz) > suggests unconvincing test results. > Actually, _most_ recent acitivi

Re: [HACKERS] WIP: store additional info in GIN index

2013-03-03 Thread Craig Ringer
The GIN changes don't seem to have progressed in some time, and some of the most recent activity (http://archives.postgresql.org/message-id/50bff89a.7080...@fuzzy.cz) suggests unconvincing test results. Is this work considered to be a dead-end - a good idea that didn't work out in practice? Or do

Re: [HACKERS] WIP: store additional info in GIN index

2012-12-23 Thread Tomas Vondra
Hi! On 22.12.2012 17:15, Alexander Korotkov wrote: > I'm not saying this is a perfect benchmark, but the differences (of > querying) are pretty huge. Not sure where this difference comes from, > but it seems to be quite consistent (I usually get +-10% results, which > is negligible

Re: [HACKERS] WIP: store additional info in GIN index

2012-12-22 Thread Alexander Korotkov
Hi! On Thu, Dec 6, 2012 at 5:44 AM, Tomas Vondra wrote: > Then I've run a simple benchmarking script, and the results are not as > good as I expected, actually I'm getting much worse performance than > with the original GIN index. > > The following table contains the time of loading the data (no

Re: [HACKERS] WIP: store additional info in GIN index

2012-12-05 Thread Tomas Vondra
On 5.12.2012 09:10, Alexander Korotkov wrote: > On Wed, Dec 5, 2012 at 1:56 AM, Tomas Vondra > wrote: > > Thanks for bug report. It is fixed in the attached patch. Hi, I gave it another try and this time it went fine - I didn't get any segfault when loading the data, which

Re: [HACKERS] WIP: store additional info in GIN index

2012-12-05 Thread Alexander Korotkov
On Wed, Dec 5, 2012 at 1:56 AM, Tomas Vondra wrote: > On 4.12.2012 20:12, Alexander Korotkov wrote: > > Hi! > > > > On Sun, Dec 2, 2012 at 5:02 AM, Tomas Vondra > > wrote: > > > > I've tried to apply the patch with the current HEAD, but I'm getting > > segfaults whe

Re: [HACKERS] WIP: store additional info in GIN index

2012-12-04 Thread Bruce Momjian
On Tue, Dec 4, 2012 at 05:35:27PM -0500, Tom Lane wrote: > Alvaro Herrera writes: > > Maybe we can mark GIN indexes as invalid after pg_upgrade somehow, so > > that they require reindex in the new cluster before they can be used for > > queries or index updates. > > Bumping the version number in

Re: [HACKERS] WIP: store additional info in GIN index

2012-12-04 Thread Tom Lane
Alvaro Herrera writes: > Maybe we can mark GIN indexes as invalid after pg_upgrade somehow, so > that they require reindex in the new cluster before they can be used for > queries or index updates. Bumping the version number in the GIN metapage would be sufficient. regard

Re: [HACKERS] WIP: store additional info in GIN index

2012-12-04 Thread Tomas Vondra
On 4.12.2012 20:12, Alexander Korotkov wrote: > Hi! > > On Sun, Dec 2, 2012 at 5:02 AM, Tomas Vondra > wrote: > > I've tried to apply the patch with the current HEAD, but I'm getting > segfaults whenever VACUUM runs (either called directly or from autovac > work

Re: [HACKERS] WIP: store additional info in GIN index

2012-12-04 Thread Bruce Momjian
On Tue, Dec 4, 2012 at 05:35:24PM -0300, Alvaro Herrera wrote: > > This means to have two versions of code which deals with posting trees and > > lists. For me it seems unlikely we have resources for maintenance of this. > > Witness how GIN has gone with unfixed bugs for months, even though > pat

Re: [HACKERS] WIP: store additional info in GIN index

2012-12-04 Thread Alvaro Herrera
Alexander Korotkov escribió: > On Tue, Dec 4, 2012 at 9:34 PM, Robert Haas wrote: > > > On Sun, Nov 18, 2012 at 4:54 PM, Alexander Korotkov > > wrote: > > > Patch completely changes storage in posting lists and leaf pages of > > posting > > > trees. It uses varbyte encoding for BlockNumber and O

Re: [HACKERS] WIP: store additional info in GIN index

2012-12-04 Thread Alexander Korotkov
On Tue, Dec 4, 2012 at 9:34 PM, Robert Haas wrote: > On Sun, Nov 18, 2012 at 4:54 PM, Alexander Korotkov > wrote: > > Patch completely changes storage in posting lists and leaf pages of > posting > > trees. It uses varbyte encoding for BlockNumber and OffsetNumber. > > BlockNumber are stored inc

Re: [HACKERS] WIP: store additional info in GIN index

2012-12-04 Thread Alexander Korotkov
Hi! On Sun, Dec 2, 2012 at 5:02 AM, Tomas Vondra wrote: > I've tried to apply the patch with the current HEAD, but I'm getting > segfaults whenever VACUUM runs (either called directly or from autovac > workers). > > The patch applied cleanly against 9b3ac49e and needed a minor fix when > applied

Re: [HACKERS] WIP: store additional info in GIN index

2012-12-04 Thread Andres Freund
On 2012-12-04 10:04:03 -0800, Josh Berkus wrote: > On 12/4/12 9:34 AM, Robert Haas wrote: > > On Sun, Nov 18, 2012 at 4:54 PM, Alexander Korotkov > > wrote: > >> Patch completely changes storage in posting lists and leaf pages of posting > >> trees. It uses varbyte encoding for BlockNumber and Off

Re: [HACKERS] WIP: store additional info in GIN index

2012-12-04 Thread Josh Berkus
On 12/4/12 9:34 AM, Robert Haas wrote: > On Sun, Nov 18, 2012 at 4:54 PM, Alexander Korotkov > wrote: >> Patch completely changes storage in posting lists and leaf pages of posting >> trees. It uses varbyte encoding for BlockNumber and OffsetNumber. >> BlockNumber are stored incremental in page. A

Re: [HACKERS] WIP: store additional info in GIN index

2012-12-04 Thread Robert Haas
On Sun, Nov 18, 2012 at 4:54 PM, Alexander Korotkov wrote: > Patch completely changes storage in posting lists and leaf pages of posting > trees. It uses varbyte encoding for BlockNumber and OffsetNumber. > BlockNumber are stored incremental in page. Additionally one bit of > OffsetNumber is reser

Re: [HACKERS] WIP: store additional info in GIN index

2012-12-01 Thread Tomas Vondra
On 18.11.2012 22:54, Alexander Korotkov wrote: > Hackers, > > Patch completely changes storage in posting lists and leaf pages of > posting trees. It uses varbyte encoding for BlockNumber and > OffsetNumber. BlockNumber are stored incremental in page. Additionally > one bit of OffsetNumber is rese

[HACKERS] WIP: store additional info in GIN index

2012-11-18 Thread Alexander Korotkov
Hackers, Attached patch enables GIN to store additional information with item pointers in posting lists and trees. Such additional information could be positions of words, positions of trigrams, lengths of arrays and so on. This is the first and most huge patch of serie of GIN improvements which w