It's now the morning of the second day of FOSDEM, and my third day in Brussels. Right now, we have a talk from Gabriele Bartolini about the Italian PUG. Yesterday was packed with several good talks from a lot of different people, and a really packed booth. I've never seen a pg booth that had this many people or this much stuff (flyers, folders, t-shirts, mugs, those lovely elephants, pins, etc etc). Lots of visitors, and lots of interesting questions.
Outside the conference, I arrived on thursday evening, and was picked up at the airport by Susanne (thanks a bunch!). Met up with Bruce and Luke Momjian at the hotel, and went to dinner with a couple of MySQL guys (one which happened to be Susannes boyfriend, which I guess explains some of the connection there..). On friday we did some touristing around Brussels, and I took the chance to do some pg work while Bruce fell asleep more or less in the middle of a sentence - I guess we blame jetlag... Susanne was off to pick up Stefan at the airport, but since he gave her his departure time instead of arrival time, the day ended up being a bit delayed...
For the evening, we met up with a whole lot of other people (Dave, Greg, Heikki, Simon, Gevik, David, and several more) for our own beer event since the experience from previous years says that the big FOSDEM one is just too crowded. Had dinner at the same place.... Food and drink was good, but it was a bit of a boring atmosphere. On the other hand, the company was good and the discussions interesting...
Second day had us setting up our two tables in the booth in the morning, stacking it with all our nice merchandise. We sold a lot of things during the whole day, and pretty much ran out of the cool plush elephants. Most of the time the booth was packed with both PostgreSQL people manning it and visitors interested in talking to the PostgreSQL people, reading through our flyers, or buying our stuff.
In the afternoon, the talks in the devroom started. David Fetter was off with the keynote which was good, but some of it was lost because he never got the projector working properly from his laptop. This was followed up by an interesting talk by Stefan on the PostgreSQL infrastructure (ok, I knew most of that, but it was still good!). Simon gave a talk on some of the new performance features in 8.3, and he really could've used much more time to go through more of them - there are just so many things to talk about in that area. Bruce finished off the days pg talks by a repeat of his "Decade of PostgreSQL" talk that he did in Toronto - always a good talk.
In the evening we filed out to a restaurant fairly close to the hotel - the biggest problem was finding one that actually had space for all of us. We ended up at an Italian place that actually also served some Belgian specialties. Some people went home from that, and the rest of us ended up having some beers and chatting about PostgreSQL, The World and Everything.
Now off to focus on todays talks. And get ready for my own, which is later on during the day.
...is not over yet. There are still things to go. But I've finished off my two talks (one on "win32 development in a unix project" (no, there was no code at all in this) and one on "building search.postgresql.org" (very little code in this one). I think they went reasonably well, we'll see what others say when I check around some more...
Not quite as many people came to our booth today, at least not the time I was there, but there were still a lot of people. We had more talks instead, so I'd still classify the day as a big success.
Gabriele kicked off in the morning with an interesting talk, that unfortunately didn't have as many visitors as it deserved, because it was early in the morning (9AM).
Bruce had a talk about the PostgreSQL future, which probably wasn't what people expected. I think most people were expecting a roadmap for 8.4, which isn't at all what we received - but it was still a good talk.
Susanne did her encoding talk and a talk on how people can contribute to the PostgreSQL community. Unfortunately, I missed both of those. David also had a talk about DBI-link that I also missed - some people had to man the booth, and since there was no scheduled time for lunch some talks simply had to be skipped.
Simon has had one talk, focusing on enterprise features that he will be working on to get into the next version(s) of PostgreSQL, and he just started his second talk which is on the topic of high availability. He also arranged a short "roundtable discussion" with the BSD folks on what they could do in the OS to help PostgreSQL. Nothing concrete from such a short meeting of course, but hopefully it laid some foundation for more work in the future.
We had an auction for one of the big elephants, which really didn't end up where we expected it - because we had only a single bidder. Probably the timing could've been a lot better since there were talks going on - we need to think about that one next time.
In about half an hour it's time for the vote for the first board of directors for the new PostgreSQL Europe organization. Going to be interesting to see how that works out, and of course what the end result of it is.
Several completely independent people have asked me this in email already, so I figured it's about time for a blog post on the topic of PostgreSQL on 64-bit versions of Windows.
First for the simple answer: No, there is no 64-bit version of PostgreSQL for Windows. PostgreSQL has supported 64-bit environments on Unix for many years (long before we had x64 to make it available for wintel machines), but there is no Win64 port. Yet. And given the way that PostgreSQL is developed, there is no firm date for when this will be available.
Until then, though, the 32-bit version of PostgreSQL runs perfectly fine on 64-bit Windows. And doing so may actually be a better idea in some cases than a native 64-bit version. But most importantly, the architecture of PostgreSQL makes the requirement for a native 64-bit version much less than for many other competing databases. Particularly:
- PostgreSQL relies on the operating system to do much of the caching of data. Since the main limitation of running a 32-bit process is the total amount of memory being addressable, a system that is based on the database engine doing all the caching cannot address all the memory on a system with say 16Gb memory. For PostgreSQL, we leave much of that caching to the operating system which can address it, as long as the OS is 64-bit.
- PostgreSQL uses a multi-process architecture, not multi-thread. I'm not going to get into the discussion about which is better, just summarize by saying there are good and bad points to both of them. In this case it's a good thing. In a multi-threaded database server, all client connections share the memory space, and are again limited by the total addressable memory region. With PostgreSQL, you could easily have more than 1Gb per backend (if necessary) without running out of memory, which significantly pushes away the point when you need to go to 64-bit.
So why would it ever be better to run a 32-bit version than a 64-bit version? Simply because a 64-bit version is more memory-hungry. Every pointer used in the system takes twice as much space. Every int takes twice as much space. And if you're not actually going be able to cash in on the benefits (by addressing loads of memory, for example), it's a very real cost.
I'm well aware that this is a fairly simplistic view on the 32- vs 64-bit situation in general, but it should be enough to explain why 32-bit PostgreSQL on 64-bit Windows isn't such a bad idea after all. And of course, this only applies if you are using an x64 version of Windows - with Itanium it's a whole different story.
Despite a couple of people pre-announcing it, PostgreSQL 8.3 was actually fully announced and released about 40 minutes ago. Announce email should be going out, and the press release is available. Instead of just reiterating the new features, I'll point you to the list of cool new stuff. Or check the release notes.
If you haven't tested your apps with it during beta or RC time, now is the perfect time to do it!
There, I just had to get this one in before
Devrim did. Ha! I win this time!
Latest comments