Amit Langote
amitl****@gmail*****
2013年 11月 1日 (金) 01:32:15 JST
On Fri, Nov 1, 2013 at 1:24 AM, Fujii Masao <masao****@gmail*****> wrote: > On Thu, Oct 31, 2013 at 11:45 PM, Amit Langote <amitl****@gmail*****> wrote: >> On 10/31/13, Fujii Masao <masao****@gmail*****> wrote: >>> On Wed, Oct 30, 2013 at 6:43 PM, Amit Langote <amitl****@gmail*****> >>> wrote: >>>> Hi, >>>> >>>> To be able to use pg_trgm's GiST related code, I ma trying to >>>> understand its code. >>>> >>>> I also tried to make a pg_bigm version by suitable (as far as I could) >>>> changes, but, for a large column it failed with - >>>> >>>> ERROR: index row requires 12544 bytes, maximum size is 8191 >>> >>> Yeah, that's one of problems of GiST index.. >>> >>>> I think the makesign() which produces GiST signature would be the one >>>> which needs the main overhaul, does that make sense? >>> >>> TBH, I've never read the source code of GiST part of pg_trgm. >>> So I'm not sure if that makes sense or not. I was guessing that >>> there are lots of things we need to change, to make pg_bigm >>> support GiST index. >>> >> >> I think this a good chance for me study both GiST and pg_bigm >> internals. I'd like give it a shot in my free time. > > Go for it! > > But I'm a bit wondering if there are any users who want to use > pg_bigm with GiST index... > I too had a similar doubt after going over the use cases of GiST in general. Do you think it's a good idea to spend time on it for pg_bigm (I would do it just for exploration of its workings though)? -- Amit
Fork