Re: Bitmap Indexes patch (was Re: Bitmap Indexes: request for feedback)

From: Gianni Ciolli <gianni(dot)ciolli(at)2ndquadrant(dot)it>
To: Greg Stark <stark(at)enterprisedb(dot)com>, Vladimir Sitnikov <sitnikov(dot)vladimir(at)gmail(dot)com>
Cc: PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org>, Gabriele Bartolini <gabriele(dot)bartolini(at)2ndquadrant(dot)it>, Simon Riggs <simon(at)2ndquadrant(dot)com>
Subject: Re: Bitmap Indexes patch (was Re: Bitmap Indexes: request for feedback)
Date: 2008-11-04 14:00:59
Message-ID: 20081104140059.GC3317@fune
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Mon, Nov 03, 2008 at 04:53:28PM -0700, Vladimir Sitnikov wrote:
> I wish to focus on the performance aspect of the patch, however, it turned
> out there are major issues with functionality: the index stores wrong tids
> inside :(
> I really would love to fix that issue and have a chance to validate the
> performance. Unfortunately, I have spent more than a day with almost void
> success.

This can be helpful for us to explain one of the two open issues that
we mentioned at submission time (meanwhile we have just fixed the
other one):
On Sat, Nov 01, 2008 at 01:01:54AM +0100, Gianni Ciolli wrote:
> * Our workaround for HOT tuples has still one bug; we are currently
> working on it and we expect to fix it soon. This bug can be
> reproduced by looking at the "rows" column of the performance test.

As for the other problem:

On Mon, Nov 03, 2008 at 05:37:24PM -0500, Greg Stark wrote:
> There are a lot of comments in the code which imply that vacuuming is
> not implemented but in fact from what I can see it is -- sort of. It
> does rewrite the bitmap in bmbulkdelete but it doesn't have to rebuild
> the index from scratch. Are the comments out of date or am i
> misunderstanding them or the code? How complete is the vacuum
> implementation?

This morning I looked at that part of the code, and I found that
indeed the vacuum implementation has a lack that we didn't
notice. After refactoring we had made some tests which suggested that
vacuum was working, but now I realize that in the hurry we missed
something.

Now, the point is that this VACUUM problem might need more work than
we expected, and that it might just be too much work for a review
phase; so, despite of the interest that showed up regarding this
feature, I will understand if the decision will be to withdraw the
patch from this Commitfest and postpone it for the next development
phase.

Thank you to everyone for your remarks,

Dr. Gianni Ciolli - 2ndQuadrant Italia
PostgreSQL Training, Services and Support
gianni(dot)ciolli(at)2ndquadrant(dot)it | www.2ndquadrant.it

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Zdenek Kotala 2008-11-04 14:06:38 Re: [WIP] In-place upgrade
Previous Message Fujii Masao 2008-11-04 13:59:58 Re: Synchronous replication patch v1