On Thu, Jul 10, 2014 at 4:20 PM, Alvaro Herrera
<alvherre@2ndquadrant.com> wrote:
> Claudio Freire wrote:
>> On Wed, Jul 9, 2014 at 6:16 PM, Alvaro Herrera <alvherre@2ndquadrant.com> wrote:
>> > Another thing I noticed is that version 8 of the patch blindly believed
>> > the "pages_per_range" declared in catalogs. This meant that if somebody
>> > did "alter index foo set pages_per_range=123" the index would
>> > immediately break (i.e. return corrupted results when queried). I have
>> > fixed this by storing the pages_per_range value used to construct the
>> > index in the metapage. Now if you do the ALTER INDEX thing, the new
>> > value is only used when the index is recreated by REINDEX.
>>
>> This seems a lot like parameterizing.
>
> I don't understand what that means -- care to elaborate?
We've been talking about bloom filters, and how their shape differs
according to the parameters of the bloom filter (number of hashes,
hash type, etc).
But after seeing this case of pages_per_range, I noticed it's an
effective-enough mechanism. Like:
CREATE INDEX ix_blah ON some_table USING bloom (somecol) WITH (BLOOM_HASHES=15, BLOOM_BUCKETS=1024,
PAGES_PER_RANGE=64);
Marking as read-only is ok, or emitting a NOTICE so that if anyone
changes those parameters that change the shape of the index, they know
it needs a rebuild would be OK too. Both mechanisms work for me.