Fujii Masao
masao****@gmail*****
2013年 11月 22日 (金) 12:51:12 JST
On Fri, Nov 22, 2013 at 9:28 AM, Amit Langote <amitl****@gmail*****> wrote: > > > On Friday, November 22, 2013, Fujii Masao wrote: >> >> On Thu, Nov 21, 2013 at 11:36 PM, Amit Langote <amitl****@gmail*****> >> wrote: >> > >> > >> > On Thursday, November 21, 2013, Fujii Masao wrote: >> >> >> >> On Thu, Nov 21, 2013 at 12:46 AM, Amit Langote >> >> <amitl****@gmail*****> >> >> wrote: >> >> > >> >> > >> >> > On Wednesday, November 20, 2013, Beena Emerson wrote: >> >> >> >> >> >> Hello, >> >> >> >> >> >> The attached patch adds detailed information about bigm_similarity >> >> >> function >> >> >> >> >> > >> >> > About the gin pending stats, is there any relation with FASTUPDATE >> >> > parameter >> >> > worth mentioning like it's being turned on or off? >> >> >> >> Sorry. I could not understand the point. Could you elaborate this? >> > >> > >> > If FASTUPDATE is turned off would there be a pending list? I am under >> > an >> > impression that there wouldn't be any; is that so? >> >> Yes, GIN index created with FASTUPDATE disabled doesn't have a pending >> list. >> So, in this case, pg_gin_pending_stats' pages and tuples columns always >> return >> 0. This is the same behavior as pgstatginindex() function which is >> provided by >> pgstattuple contrib module. >> > > Ah, I see. If users would already be aware of such behavior then I guess > it's not necessary to stress FASTUPDATE in documentation. So, you're thinking that we should document something like the following? Note that pg_gin_pending_stats' pages and tuples columns return 0 if its argument is the GIN index built with FASTUPDATE option disabled because it doesn't have a pending list. Please see <link to the proper PostgreSQL document> for the details of FASTUPDATE option. I don't have any strong opinion about this. If you think this is important, could you implement the patch? I will commit that! Regards, -- Fujii Masao
Fork