Tom Lane wrote: > Bruce Momjian <[EMAIL PROTECTED]> writes: > > Tom Lane wrote: > >> Where was it posted anyway? > > > Found it: > > > > > http://groups.google.com/groups?hl=en&lr=&ie=UTF-8&selm=200312010450.hB14ovH16330%40candle.pha.pa.us&rnum=8 > > Thanks. The original patch is much older than I thought --- I was > looking in the November/December part of the archives. > > > Personally, because frequently accessed duplicates appear more forward > > in the duplicate index, I think the sorting is only valuable when > > creating a new index. > > Yes, and that's what this does. Looking back, the original discussion > got a little confused because the TODO item about "order duplicate index > entries by tid" got brought into the mix. Actually this patch has > nothing to do with that, because it only acts during btree creation not > during index updates. > > On inspection I have no problem with the patch, only with the comments ;-) > If you like I'll revise the comments and apply.
Great. Seems harmless and he showed good performance with it. I agree the discussion got confused, and that is why I kept it in my mailbox to revisit. -- Bruce Momjian | http://candle.pha.pa.us [EMAIL PROTECTED] | (610) 359-1001 + If your life is a hard drive, | 13 Roberts Road + Christ can be your backup. | Newtown Square, Pennsylvania 19073 ---------------------------(end of broadcast)--------------------------- TIP 9: the planner will ignore your desire to choose an index scan if your joining column's datatypes do not match
