The Linux Journal announced the Winners of the 2003 Editors Choice awards.

Of course, the best database is PostgreSQL.

Contributors: snpe snpe at snpe.co.yu, elein at varlena.com

Our Birds of A Feather Session on Wednesday evening was the first time I got to see all of the PostgreSQL folks all together at the same time. In spite of not setting up the meeting before the conference about thirty people attended and the discussion was lively. Someone else will have to tell the tale of the late night dinners.

The particular feedback that hit home to me (others' lists may vary) was our lack of organized marketing, lack of visibility, and the chaotic nature of our websites. None of these complaints were a surprise to anyone. The advocacy group has made some significant decisions with regards to rectifying these items.

Perhaps not everyone knows that three items have happened recently to further our marketing efforts. First, Josh Berkus was drafted into the Core Development Group. Josh's strength is in organization and marketing (and performance as you saw in the previous issue). Under Josh's watch, efforts to fund a marketing have begun. Marc Fournier has also announced that part of the proceeds from banner ad on the postgresql.org web pages will be contributed to the advocacy funds. We are also planning on having a booth, a presence, at Comdex. In addition to these items, volunteers are working on brochures, CD covers, etc. so that we have some marketing collateral. It is good to know we are on the right track.

Another part of the discussion at the birds of a feather session was the native PC port which is delayed until after the release of 7.4. Discussion about why the port is and is not important, particularly in light of marketing and visibility issues, were both given fair air time.

PostgreSQL people were not particularly visible in the general milieu of the conference. We were there, but not seen too much. Several of the talks in the conference and discussions in our birds of a feather session pointed out that open source projects need to establish relationships with other open source organizations. I know that some of us are members of more than one open source group, but for the most part, we remain fairly isolated. Talk to each other. Reason things out. This is open source so steal ideas that work! For example, we need to have more visible, vocal and active user groups across the world, including the United States. And perl-mongers is quite successful. What advice can we get from them?

The establishment of relationships between people and groups was raised in many different contexts at the conference. PostgreSQL, the database, particularly plays well with others. Is there something in this that we can use to promote our favorite database?

One mistake that was made at least once was to bad mouth the perceived competition. Venting one's frustration is one thing, but, particularly with our minimal visibility, we must make a good impression. Besides, everyone knows we are technically more mature and can beat the pants off any perceived competitor. So there is no need to bully them. I am sure I have been guilty of this, too. But restraint of tongue and pen (or keyboard) is what shows strength. Ms. Manners Rules!

For us, I believe, the point is just good technology. The highest complement paid to any software or hardware is It just works! Databases are complex programs to solve the complex problems of managed data storage and retrieval. And PostgreSQL is rich with features to deal with the complexities of the problem set. We cannot pretend to be simple. But PostgreSQL has additional maturity, stability and conformance to standards, so I believe this goal is within our grasp:

Other attendees: If you have comments about the conference send them to elein@varlena.com and they will be added to these.

Discussion of the items raised here is appropriate to the PostgreSQL Advocacy Mailing list.

Editor: elein at varlena.com

Joe Conway writes:

I have no facts or figures to back this up, but it "felt" to me like the PostgreSQL community was less visible (in terms of numbers of presentations) and less attended than last year. The presentations and tutorials delivered were very good, but there just weren't enough of them.

I would have like to have seen PostgreSQL tutorials on both Monday and Tuesday (instead of just Monday), and Postgres presentations on all three days of sessions (instead of just Thursday).

I think part of the problem was that Bruce got very late notice from O'Reilly regarding submissions this year. Next year we should get the word out far and wide and early so that we get more people stepping up to speak. We should also make a concerted effort to get PostgreSQL on O'Reilly's radar.

Greg Sabino Mullane writes:

Here are my quick* notes about OSCON from a PostgreSQL perspective.
 * You should see my long notes. :)

It was great finally getting to meet so many of them names I have only seen on the Internet (mostly the mailing lists) and get to interact with people. Sometimes we even managed to talk about non-Pg/non-technical things. But not very often. :)

The Postgres presentations were all very good. I went to some of the MySQL presentations, and they were not that good. Maybe they should spend some of that $19 million on public speaking classes. I particularly enjoyed Tom Lane's talk, as that one was the closest to mine - SQL tuning - and gave a nice glimpse into the changes for 7.4. Very impressive stuff. Even Marty from MySQL was at that talk, asking questions. I am constantly amazed at some of the smart things Postgres does behind the scenes.

Someone made the argument that we don't need to focus on the Win32 port right now, as the Microsoft share of the market is declining. This seemed to be the minority view, especially in light of the fact that we are very close to being done. My own take on this is that we are the only database that I can think of that does *not* have a native Win32 port. If it is good enough for Oracle...

Someone else mentioned that they would often go to sourceforge while looking for a project to help them out and would end up converting the app to Postgres, as most of them are only for MySQL. Although this was a side comment to something else, it got me to thinking: why not have another part of Postgres that would coordinate converting all the apps in the world to run with PostgreSQL? I envision a page where people could add projects that need to be converted to a list. Other people could do the actual converting, perhaps with a final checkoff by a reviewer, and the project could be marked as "done." We would not only automatically generate a list of applications that run PostgreSQL, but would have another list that would show the status of other apps. Outside users could even request to have a particular app ported. I see this as being a very important part of the PostgreSQL project, similar in importance to the web, source, and advocacy projects. When I get some time, I'll probably start a prototype.

Another thing mentioned was the lack of visibility at OSCON. Although we had some very good, high-quality presentations, I feel that our visibility was absolutely terrible at OSCON. We were off hiding in a corner while MySQL was out there shouting "Here I am!". From the t-shirt given to each participant (MySQL, *not* O'Reilly like in previous years) to Tim O'Reilly's reinforcement of the terrible LAMP acronym in his keynote, to the lack of any mention of Postgres in the "hallway" or when discussing other projects, Postgres just seemed to be off the radar. Our only redeeming feature was that we had our own track. Even with that, we should have had more presentations. I have some ideas for next year's, and I think we need to hit the conference early and hard with submissions. I especially would like to see presentations that tie Postgres into other projects and/or applications. I would also like to see more that target the "unknown" user - people who have heard of Postgres, but are new to it or are still evaluating it. Sometimes I feel we preach to the choir at the expense of new converts.

A radical idea about reorganizing underneath the Apache project was bravely tossed out to the group. I don't think that would ever happen (but spent some time debating the idea with the originator the next day), but something like that is certainly needed as far as consolidating the PostgreSQL project into a single legal entity that can do things like receive funds, send people to conferences, publish materials, coordinate contacts, etc. Almost a combination of the core group and the advocacy group. Apache has an interesting model: a "core" business (actually a foundation, which may be the best for Postgres), and a bunch of "interest groups" that handle one project. In our case, I could easily see a main foundation that includes a technical group (source code), an advocacy group, a web site group (also FTP/DNS/etc), and an application group (as mentioned above, whose goal is to port other projects). I have more ideas about all of this, but in the interest of space, I shall move on.

Someone threw out the ol' "let's ignore MySQL, and focus on DB2" argument. (Feel free to replace "DB2" with "Microsoft SQL Server" or "Oracle" of course). I think this is a big mistake. As I pointed out to the group, we need to be concerned with /all/ the other players, not just the large commercial ones. Someone else (sorry, do not know who) pointed out that while MySQL might have been a "toy" 18 months ago, they are not now, and are moving forward at a significant speed. I was fairly surprised to hear the "ignore MySQL" argument at an open-source conference of all places. Winning the mindshare of the open-source community should be one of our highest goals. And right now, MySQL has that mindshare firmly in their hands. Underestimating the importance of that mindshare, and underestimating MySQL, are two things that the Postgres project cannot afford to do. Most people at the BOF seemed to agree that MySQL is still a concern.

As mentioned before, the importance and timing of future features was mentioned. Things on the radar include replication, point-in-time recovery (PITR), two-phase commit (2PC), nested transaction, and the Win32 port. Some things (replication) are much closer than others (nested transactions). All of them are needed, but after the BOF I was starting to wonder: what now? What's the next Big Thing after we get those five done? (Using "we" in the "I ain't written a single line of code to help any of those five out" of course! :) I am sure that replication will still be a monster, as there are many ways to do it and it can can very complex. Still, it will soon be one less thing the MySQL folks can beat us up about once we have a working, built-in solution.

We also talked about the problem of getting companies to state that they are using Postgres. It's a chicken-and-egg problem: nobody will admit that they are using Postgres until lots of other companies are publicly admitting that they are using Postgres. Major thanks to Andrew Sullivan for starting that egg rolling by extracting enough information from other companies to enable him to fight hard to get Postgres as the backend for the .info and .org domains. Someone else (again, cannot remember many names) suggested that we should be recording all companies that use Postgres, even if they are not large. Someone else asked about the case studies: as far as anyone knows, we have bunch, but they are not on the website yet.

There is lot more to say, but overall it was a great experience, not just from a PostgreSQL perspective, but from a general technical one as well. Postgres has a very strong community, but it needs some visibility and legitimacy. I've got some great t-shirts idea for next year...

A late entry, David Fetter Writes:

The "LAMP" concept is a serious obstacle to PostgreSQL adoption and development. This requires a substantial marketing effort. How big is MySQL's purse? To what extent is O'Reilly politically committed to it?

I realized that what PostgreSQL really needs is an aggressive and well-funded marketing effort.

I've found that it is unproductive to pretend that a debate is happening when the other side is not engaging in a debate. ...[other projects] have been consistently engaging in deception, misinformation and outright lies... To miss that point, publicly or privately, is to lose the fight.

As promised, here is the talk that I gave at OSCON. There are three parts, the script notes, the slides and the source code for the examples.

Talk on Running Aggregates using PostgreSQL and plpython.
© A. Elein Mustain 2003
Presented July 2003 at O'Reilly Open Source Conference, Portland, OR

• Slides
• Source Code
• Script

  Caveat: These are my notes on this talk. They are not necessarily complete nor are they a definitive record of any live talk, real or imagined.

  Modifications to content of the original script based on new information are noted in italics.

As always, comments and questions are welcome.

Contributors: elein at varlena.com