maria-developers team mailing list archive

[Sergei Golubchik <serg@xxxxxxxxxxx>]

Hi, Sachin!

On Apr 19, Sachin Setia wrote:
> Hello Sergei!
> I am sorry , I pressed Ctrl + Enter so first message is an incomplete
> message Actually first I was thinking instead of storing hash as b+tree
> store it in hash table But then i come to know that we can have range
> query in blob fields If we use hash table then range query will take
> O(n) time. I was reading mysql manual then i come across this term this
> will monitor mysql query and if it finds that there is lot of query
> using in or equal then it will build hash index from b+tree index.I want
> to implement same thing in myisam

Let's rather try to do it on the upper layer, in sql/
And then I'd be very interested to see if your hash index idea could be
used to speed up SELECT DISTINCT. Long unique constraints in normal
tables are not that interesting - a table might have many unique
constraints, a table might have occasional inserts, mixed with selects,
updates, deletes, and accesses to other tables. But a temporary table
for SELECT DISTINCT - it has only one unique constraint, it gets lots of
inserts/updates in a short period of time - it should be a perfect
target for your hash index optimization.

Regards,
Sergei
Chief Architect MariaDB
and security@xxxxxxxxxxx