Release Notes: Major Changes in 8.2

Simon Riggs wrote:

Zero administration overhead now possible (Alvaro)

With autovacuum enabled, all required vacuuming will now take place
without administrator intervention enabling wider distribution of
embedded databases.

This was true for 8.1 already, no?

Regards,
Andreas

Andreas Pflug wrote:

Simon Riggs wrote:

Zero administration overhead now possible (Alvaro)

With autovacuum enabled, all required vacuuming will now take place
without administrator intervention enabling wider distribution of
embedded databases.

This was true for 8.1 already, no?

No. 8.1 did not have it turned on by default.

Sincerely,

Joshua D. Drake

Regards,
Andreas

---------------------------(end of broadcast)---------------------------
TIP 6: explain analyze is your friend

--

=== The PostgreSQL Company: Command Prompt, Inc. ===
Sales/Support: +1.503.667.4564 || 24x7/Emergency: +1.800.492.2240
Providing the most comprehensive PostgreSQL solutions since 1997
http://www.commandprompt.com/

On Wed, 2006-09-20 at 18:22 +0200, Andreas Pflug wrote:

Simon Riggs wrote:

Zero administration overhead now possible (Alvaro)

With autovacuum enabled, all required vacuuming will now take place
without administrator intervention enabling wider distribution of
embedded databases.

This was true for 8.1 already, no?

Hmmm. You're correct.

Perhaps that is not a major change after all.

--
Simon Riggs
EnterpriseDB http://www.enterprisedb.com

Simon Riggs wrote:

On Wed, 2006-09-20 at 18:22 +0200, Andreas Pflug wrote:

Simon Riggs wrote:

Zero administration overhead now possible (Alvaro)

With autovacuum enabled, all required vacuuming will now take place
without administrator intervention enabling wider distribution of
embedded databases.

This was true for 8.1 already, no?

Hmmm. You're correct.

Perhaps that is not a major change after all.

What happened in 8.2 is that you no longer need database-wide vacuums,
ever (except for template databases). Not sure if that qualifies as a
major change or not.

--
Alvaro Herrera http://www.CommandPrompt.com/
PostgreSQL Replication, Consulting, Custom Development, 24x7 support

"Joshua D. Drake" <jd@commandprompt.com> writes:

No. 8.1 did not have it turned on by default.

Neither does 8.2 though.

--
Gregory Stark
EnterpriseDB http://www.enterprisedb.com

Gregory Stark wrote:

"Joshua D. Drake" <jd@commandprompt.com> writes:

No. 8.1 did not have it turned on by default.

Neither does 8.2 though.

oh... heh.

J

--

=== The PostgreSQL Company: Command Prompt, Inc. ===
Sales/Support: +1.503.667.4564 || 24x7/Emergency: +1.800.492.2240
Providing the most comprehensive PostgreSQL solutions since 1997
http://www.commandprompt.com/

Usually the major items just jump out of the release list. In this
case, nothing really jumped out, and I felt if I listed sereral, it was
going to look weak because they were not big things, so I figured I
would just go with the "broad" list.

The criteria I usually use are things that were not easy to do before.

Does the list below look good for inclusion?

I guess my point is that what we have now overwhelms people with the
number of small things we did. If you try to put a few at the top, does
it diminish it because the top things are not large?

Or perhaps we can do more broad-stroke list items, like monitoring or
performance, as listed below.

---------------------------------------------------------------------------

Simon Riggs wrote:

I'd like to include a section on Major changes in this release at the
top of the release notes, as has been done for at least the last 6 major
releases. The notes below are one stab at that, for **discussion**. I've
tried to arrange specific changes into groups...

Major changes in this release:

Improved scalability and performance on multi-processor systems (Tom,
Alvaro, Itagaki, Qingqing, Heikki)

A variety of changes improves the performance of both sequential scans
and index scans, as well as enhancing multi-processor scalability. The
advanced query optimizer has also been further enhanced, allowing
indexes and partitioning to be useful in more cases.

Improved utility and large query performance (Tom, Simon, Alon, Andreas)

Large sorts will have typical performance increases of 100-300%,
improving complex queries and creating new indexes. Loading times have
also been reduced. Large queries, data loads, upgrades and restores will
be considerably improved.

Improved monitoring and performance tuning (Tom, Bruce, Greg, Larry)

Overhead of statistics collection has been considerably reduced and new
statistics and system information is available. Better query logging
improves diagnostics and especially performance tuning. Server now
includes DTrace support. Indexes can now also be created CONCURRENTLY,
allowing application tuning without effecting server availability.

Zero administration overhead now possible (Alvaro)

With autovacuum enabled, all required vacuuming will now take place
without administrator intervention enabling wider distribution of
embedded databases.

Improved defaults and configuration (Peter, Andrew)

Installation defaults are now improved for many tunable memory
parameters and these can now be specified in kB, MB and GB.

Warm Standby Servers for High Availability (Simon, Tom)

Warm Standby servers can now be more easily configured and are
appropriate in a wider range of circumstances than previously.

Improved scalability and performance of text search: GIN and Tsearch2
(Teodor, Oleg)

New GIN indexes allow much larger text search indexes than were
previously possible. TSearch2 has been enhanced and performance has also
been greatly improved.

Enhanced DML Functionality (Jonah, Joe, Tom, Susanne, Atsushi)

INSERT/UPDATE/DELETE RETURNING and INSERT .. VALUES (), VALUES (),
VALUES () allow more efficient application designs. Enhancements to
UPDATE and DELETE allow additional constructs for clarity and ease of
use.

SQL:2003 Analytical functions (Sergey, Tom, Neil)

All statistical aggregate functions defined by SQL:2003 are now
supported and user-defined aggregates now can take multiple columns as
inputs.

--
Simon Riggs
EnterpriseDB http://www.enterprisedb.com

---------------------------(end of broadcast)---------------------------
TIP 4: Have you searched our list archives?

http://archives.postgresql.org

--
Bruce Momjian bruce@momjian.us
EnterpriseDB http://www.enterprisedb.com

+ If your life is a hard drive, Christ can be your backup. +

Bruce,

What happened to PL/pgSQL debugging? Did it die?

--
Josh Berkus
PostgreSQL @ Sun
San Francisco

Simon Riggs napsal(a):

Improved monitoring and performance tuning (Tom, Bruce, Greg, Larry)

Overhead of statistics collection has been considerably reduced and new
statistics and system information is available. Better query logging
improves diagnostics and especially performance tuning. Server now
includes DTrace support. Indexes can now also be created CONCURRENTLY,
allowing application tuning without effecting server availability.

You forgot to Robert Lor - author of DTrace support.

Zdenek

On Wed, 2006-09-20 at 23:22 -0400, Bruce Momjian wrote:

Usually the major items just jump out of the release list. In this
case, nothing really jumped out, and I felt if I listed sereral, it was
going to look weak because they were not big things, so I figured I
would just go with the "broad" list.

Look back at the 7.4 release notes as a comparison. I think 8.0 was such
a milestone release we tend to judge ourselves by that and maybe feel
like the pace has slackened. IMHO, it has accelerated. We hit the lower
hanging fruit first, so early features were major items; later items
seem smaller and less important by comparison, especially when completed
by a team rather than a few individuals.

I don't think it matters whether the new features originated as a single
patch or as a stream of smaller patches. The end result is a major
improvement in a specific area. Picking one area I'm more familiar with,
sort performance was increased over many patches by many people, but the
original objective of making a step-change in that area *has* been
achieved (even if there are some additional gains still to be had for
certain narrower use-cases).

The role of the "Major changes" section is to provide a summary for
administrators who need to understand what a new release will give them
and make a cost/benefit judgement. We want people to understand the good
work that has been done and that does involve some filtering and
summarization, and its possibly true that it is harder in this release
than others.

We need a Major changes section: People don't read the detail: sysadmins
are too busy these days. If there are no major features listed, people
will assume there are none and say "oh its just a bug fix release". If
we aren't encouraging people to upgrade, why release at all? Maybe
people only upgrade every other release - if so, we'll get all of the
8.0 upgraders.

Improving scalability in 8.1 was great. Improving it again in 8.2 is
amazing and we should tell people, even if it sounds somewhat boring
because we did it last time as well. I think: again, wow, this software
is going places. Personally, I'll be ecstatic if we can do that again
for 8.3...

Or perhaps we can do more broad-stroke list items, like monitoring or
performance, as listed below.

Whether we like my list or not, I think such a grouped list should
exist. I'm mainly seeking to persuade you on that point and would be
comfortable even if you came up with a different grouped list.

Seeing a list of names after a topic emphasises the community
development process. In some cases, there was a stated objective and
that has been achieved. In other cases there was a community-driven move
in directions maybe we didn't predict. In the latter case, surely it is
the strength of open source that evolution works so well and really does
produce noticeably major changes. The changes in monitoring and tuning
tools is an excellent example: many smaller changes making a significant
improvement.

Please vote in favour of a Major Changes section.

--
Simon Riggs
EnterpriseDB http://www.enterprisedb.com

Josh Berkus wrote:

Bruce,

What happened to PL/pgSQL debugging? Did it die?

The debuggers is going to be on pgfoundry, if it isn't there already.
The idea is that it would be loadable for 8.2, work out all the bugs,
and perhaps included in 8.3.

--
Bruce Momjian bruce@momjian.us
EnterpriseDB http://www.enterprisedb.com

+ If your life is a hard drive, Christ can be your backup. +

OK, I will work it.

---------------------------------------------------------------------------

Simon Riggs wrote:

On Wed, 2006-09-20 at 23:22 -0400, Bruce Momjian wrote:

Usually the major items just jump out of the release list. In this
case, nothing really jumped out, and I felt if I listed sereral, it was
going to look weak because they were not big things, so I figured I
would just go with the "broad" list.

Look back at the 7.4 release notes as a comparison. I think 8.0 was such
a milestone release we tend to judge ourselves by that and maybe feel
like the pace has slackened. IMHO, it has accelerated. We hit the lower
hanging fruit first, so early features were major items; later items
seem smaller and less important by comparison, especially when completed
by a team rather than a few individuals.

I don't think it matters whether the new features originated as a single
patch or as a stream of smaller patches. The end result is a major
improvement in a specific area. Picking one area I'm more familiar with,
sort performance was increased over many patches by many people, but the
original objective of making a step-change in that area *has* been
achieved (even if there are some additional gains still to be had for
certain narrower use-cases).

The role of the "Major changes" section is to provide a summary for
administrators who need to understand what a new release will give them
and make a cost/benefit judgement. We want people to understand the good
work that has been done and that does involve some filtering and
summarization, and its possibly true that it is harder in this release
than others.

We need a Major changes section: People don't read the detail: sysadmins
are too busy these days. If there are no major features listed, people
will assume there are none and say "oh its just a bug fix release". If
we aren't encouraging people to upgrade, why release at all? Maybe
people only upgrade every other release - if so, we'll get all of the
8.0 upgraders.

Improving scalability in 8.1 was great. Improving it again in 8.2 is
amazing and we should tell people, even if it sounds somewhat boring
because we did it last time as well. I think: again, wow, this software
is going places. Personally, I'll be ecstatic if we can do that again
for 8.3...

Or perhaps we can do more broad-stroke list items, like monitoring or
performance, as listed below.

Whether we like my list or not, I think such a grouped list should
exist. I'm mainly seeking to persuade you on that point and would be
comfortable even if you came up with a different grouped list.

Seeing a list of names after a topic emphasises the community
development process. In some cases, there was a stated objective and
that has been achieved. In other cases there was a community-driven move
in directions maybe we didn't predict. In the latter case, surely it is
the strength of open source that evolution works so well and really does
produce noticeably major changes. The changes in monitoring and tuning
tools is an excellent example: many smaller changes making a significant
improvement.

Please vote in favour of a Major Changes section.

--
Simon Riggs
EnterpriseDB http://www.enterprisedb.com

---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match

--
Bruce Momjian bruce@momjian.us
EnterpriseDB http://www.enterprisedb.com

+ If your life is a hard drive, Christ can be your backup. +

-----Original Message-----
From: pgsql-hackers-owner@postgresql.org
[mailto:pgsql-hackers-owner@postgresql.org] On Behalf Of Bruce Momjian
Sent: 21 September 2006 16:25
To: Josh Berkus
Cc: pgsql-hackers@postgresql.org; Simon Riggs
Subject: Re: [HACKERS] Release Notes: Major Changes in 8.2

Josh Berkus wrote:

Bruce,

What happened to PL/pgSQL debugging? Did it die?

The debuggers is going to be on pgfoundry, if it isn't there already.
The idea is that it would be loadable for 8.2, work out all the bugs,
and perhaps included in 8.3.

We've also discussed bundling the GUI with pgAdmin for 1.8 (which will
be released with 8.3) so that idea could work out nicely.

Regards, Dave.

Bruce, All:

The debuggers is going to be on pgfoundry, if it isn't there already.
The idea is that it would be loadable for 8.2, work out all the bugs,
and perhaps included in 8.3.

So, should I take this off the press list for 8.2 and save it for 8.3,
when the feature will be actually useful?

Second question: are the Advisory Locks actually a unique PostgreSQL
feature, or are these something other databases already have?

--Josh Berkus

On Thu, Sep 21, 2006 at 11:24:53AM -0400, Bruce Momjian wrote:

Josh Berkus wrote:

Bruce,

What happened to PL/pgSQL debugging? Did it die?

The debuggers is going to be on pgfoundry, if it isn't there already.
The idea is that it would be loadable for 8.2, work out all the bugs,
and perhaps included in 8.3.

But didn't we end up putting some hooks in the backend to make this
possible?

Regardless, I think we should include a section of major new
projects/developments from pgFoundry, because they ultimately make
PostgreSQL a more useful database. Maybe this list should only be in the
PR (and not the release notes)...
--
Jim Nasby jim@nasby.net
EnterpriseDB http://enterprisedb.com 512.569.9461 (cell)

Josh Berkus wrote:

Bruce, All:

The debuggers is going to be on pgfoundry, if it isn't there already.
The idea is that it would be loadable for 8.2, work out all the bugs,
and perhaps included in 8.3.

So, should I take this off the press list for 8.2 and save it for 8.3,
when the feature will be actually useful?

Yes, I think so.

Second question: are the Advisory Locks actually a unique PostgreSQL
feature, or are these something other databases already have?

Probably not unique.

--
Bruce Momjian bruce@momjian.us
EnterpriseDB http://www.enterprisedb.com

+ If your life is a hard drive, Christ can be your backup. +

bruce@momjian.us (Bruce Momjian) writes:

Josh Berkus wrote:

Bruce,

What happened to PL/pgSQL debugging? Did it die?

The debuggers is going to be on pgfoundry, if it isn't there already.
The idea is that it would be loadable for 8.2, work out all the bugs,
and perhaps included in 8.3.

If we now have the hooks in place, then it is surely worth saying so.
To then point people to pgFoundry for an add-on "debugger" application
seems pretty fair.
--
select 'cbbrowne' || '@' || 'cbbrowne.com';
http://linuxdatabases.info/info/finances.html
Rules of the Evil Overlord #133. "If I find my beautiful consort with
access to my fortress has been associating with the hero, I'll have
her executed. It's regrettable, but new consorts are easier to get
than new fortresses and maybe the next one will pay attention at the
orientation meeting." <http://www.eviloverlord.com/&gt;

Simon Riggs wrote:

SQL:2003 Analytical functions (Sergey, Tom, Neil)

All statistical aggregate functions defined by SQL:2003 are now
supported and user-defined aggregates now can take multiple columns as
inputs.

Could this be a good starting point for SQL:2003 Window functions as
now the work on SQL:2003 statistical functions are done? As
experienced postgres developers what would be your roadmap to implement
window functions?

I created a major features list for 8.2 and put it into CVS. Instead of
going into detail (meaning the item would not appear in the "Changes"
section below, I just highlighted some of the big stuff, and was
purposely vague about the details, so people just have an overview of
what is below.

Let me know how it looks.

Simon's list below looks good, but it really has a lot of details,
particuarly it goes into use-cases for many of the features, and in fact
goes into more detail that we even have in the release notes now. Is
that what people want? My concern is that if we push too much
information, it is hard to see the actual features, i.e. if we say, "we
have feature X, and it is good for Y, Z, and Q" do people remember Y and
Z and forget X?

Again, I don't want to be the person writing these release notes, so I
am looking for feedback, good or bad.

---------------------------------------------------------------------------

Simon Riggs wrote:

I'd like to include a section on Major changes in this release at the
top of the release notes, as has been done for at least the last 6 major
releases. The notes below are one stab at that, for **discussion**. I've
tried to arrange specific changes into groups...

Major changes in this release:

Improved scalability and performance on multi-processor systems (Tom,
Alvaro, Itagaki, Qingqing, Heikki)

A variety of changes improves the performance of both sequential scans
and index scans, as well as enhancing multi-processor scalability. The
advanced query optimizer has also been further enhanced, allowing
indexes and partitioning to be useful in more cases.

Improved utility and large query performance (Tom, Simon, Alon, Andreas)

Large sorts will have typical performance increases of 100-300%,
improving complex queries and creating new indexes. Loading times have
also been reduced. Large queries, data loads, upgrades and restores will
be considerably improved.

Improved monitoring and performance tuning (Tom, Bruce, Greg, Larry)

Overhead of statistics collection has been considerably reduced and new
statistics and system information is available. Better query logging
improves diagnostics and especially performance tuning. Server now
includes DTrace support. Indexes can now also be created CONCURRENTLY,
allowing application tuning without effecting server availability.

Zero administration overhead now possible (Alvaro)

With autovacuum enabled, all required vacuuming will now take place
without administrator intervention enabling wider distribution of
embedded databases.

Improved defaults and configuration (Peter, Andrew)

Installation defaults are now improved for many tunable memory
parameters and these can now be specified in kB, MB and GB.

Warm Standby Servers for High Availability (Simon, Tom)

Warm Standby servers can now be more easily configured and are
appropriate in a wider range of circumstances than previously.

Improved scalability and performance of text search: GIN and Tsearch2
(Teodor, Oleg)

New GIN indexes allow much larger text search indexes than were
previously possible. TSearch2 has been enhanced and performance has also
been greatly improved.

Enhanced DML Functionality (Jonah, Joe, Tom, Susanne, Atsushi)

INSERT/UPDATE/DELETE RETURNING and INSERT .. VALUES (), VALUES (),
VALUES () allow more efficient application designs. Enhancements to
UPDATE and DELETE allow additional constructs for clarity and ease of
use.

SQL:2003 Analytical functions (Sergey, Tom, Neil)

All statistical aggregate functions defined by SQL:2003 are now
supported and user-defined aggregates now can take multiple columns as
inputs.

--
Simon Riggs
EnterpriseDB http://www.enterprisedb.com

---------------------------(end of broadcast)---------------------------
TIP 4: Have you searched our list archives?

http://archives.postgresql.org

--
Bruce Momjian bruce@momjian.us
EnterpriseDB http://www.enterprisedb.com

+ If your life is a hard drive, Christ can be your backup. +