I had a great time at my first MAX presentation. About fifteen minutes before I was to present my video card fried. As a Macromedia technician worked on my computer to see if it was something easily remedied, I frantically looked for other options. Another presenter loaned me his laptop and my tack manager had a copy of the presentation, so we loaded up the laptop and I was good to go. When I turned the laptop on back at the session, I was presented with a very unwelcome sight–the login screen. I didn’t have the password.
So, I gave my first presentation with no slides, no examples, and no demonstrations. I was very surprised that only a handfull of people left the room, and noticed very confused looks on the people that came in late. Luckily the topic is more more conceptual than hands-on and based on questions from the attendees it seemed that the presentation went about as well as can be expected.
Near the end my track manager and some Macromedia techs came in with another backup computer and we set it up to help highlight the things that needed visuals for explanation and to show the URL where the presentation can be downloaded.
I didn’t have time this week to attend MAX in New Orleans, but tomorrow I’ll be on my way to MAX Taiwan. I’ll be giving two sessions, one on “Skinning Flash MX 2004 Components”, the other on “Flash MX 2004 Component Development Using the Version 2 Architecture”. There are some excellent Flash related sessions lined up, and a good list of speakers.
If anyone has suggestions on what to see/do in Taipei let me know, this will be my first time in Taiwan
This seemed counter-intuitive to us when we stumbled upon it, but apparently wrapping update statements inside a transaction can improve the performance of the statement.
We were working on an import script that imports a large set of new data and also updates related records in existing data. The process runs asynchronously so we weren’t concerned too much with performance–it’s a long running process updating 7,000,000+ existing records with related data from 100,000 new ones.
However, we were surprised to notice that the same queries took over 1,000 seconds to run through our application whereas they ran in about 150 seconds in SQL+. After a lot of trial an error we wrapped the application queries in a transaction and the performance improved to be the same as running the statements in SQL+.
Because of the nature of the data load we don’t really need autonomous commits and thus weren’t using transactions to start with. We were very surprised to notice the speed increase after applying transactions.