OSCON, Day 4, Thursday

| 1 Comment

Worked on getting pictures online for way too long last night, so I missed our lunch hookup with Brian Behlendorf and DirkX (too tired to remember anything). Aaargh. I even set it up, so I felt bad. But I guess it's okay as for me it was mostly just getting the other Perl Documentation guys to drink the ASF Kool-Aid.

Chip, Jon Orwant, Jarkko and Elaine were still in the lunch tent when I came over. Very entertaining as always.

I wanted to go to the NetTopBox talk; Orwant said that Carl is cool. I got distracted though, so I didn't make it. Aaron Swartz took notes. Excellent. Thanks Aaron!

I saw a bit of Stas talk on mod_perl 2.0. He had way too many slides (surprise surprise ;-) ) but it was good. He sent me the chapter from the book it's based on. Later I met Linda Mui (finally) who told that their book is going into formal tech review soon, very cool.

Then I saw most of Dirk's "writing apache 2.0 modules" talk. It seemed pretty useful, but I knew most of it. So I moved to catched the last bits of Allison Randals talk; it seemed really cool. People were suggesting that she should do it next year in a bigger room. It was fun. Larry, Damian, Chip and everyone were there commenting and cracking jokes.

We had a #perl photoshoot with Julian. Lots of fun. Julian liked it too. Quote: My ego has never been so stroked!
.

Shortly after I took photos of mjd laying on the floor with the octopus and a GNU t-shirt. Tim Bunce helped us make it look authentic.

Now I'm in Perrin Harkin's "shared data with mod_perl" talk. He spend forever getting to the pretty graphs. Useful stuff, but kinda information overload for my attentionspan. All the benchmarks, and that he looked at pretty much all the standard modules to do this is really cool though.

BerkeleyDB; Tie::TextDir and IPC::MM seems to be the fas "shared data" modules. Another interesting, but known, point is that shared memory isn't as fast as you would think and the file system is faster than you think. At ValueClick Graham and I once made a "shared storage thing" that was all simple flat files in a hashed directory structure. That was by far the fas we could get, and we had tried quite a few different things. (Including a very sophisticated BerkeleyDB thing with Semaphore locks).

Now he is telling about all the problems with the . As I said in my talk: Remember kids, always do your own benchmarks.

Perrin's 45 minutes went by really fast. Good sign; I usually get easily bored in talks.

It seems like the 802.11 network is much better anywhere not close to the speakers lounge thing.... Hmmmn...

There are about three thousand BoF's tonight I want to attend. Help!
DBI Bof; p5ee bof; Writing for Ora bof; Weblog bof; Foundation Bof; Aron Wall's Self-Modifying & OS Card Game; Something Interesting Dominus Is Working On. And then there is of course the Free Beer^W^WLAN Party thing.

1 Comment

At ValueClick Graham and I once made a "shared storage thing" that was all simple flat files in a hashed directory structure. That was by far the fastest we could get, and we had tried quite a few different things.

Any chance you could make a CPAN module out of this?

--David

Leave a comment

About this Entry

This page contains a single entry by Ask Bjørn Hansen published on July 25, 2002 4:59 PM.

Find recent content on the main index or look in the archives to find all content.

Pages

OpenID accepted here Learn more about OpenID
Powered by Movable Type 4.33-en
/* bf */