You write a prompt, hit enter, and watch the console light up green. Code pours out. It feels like magic. Then you try to maintain the thing a week later, and you realize your shiny new AI agent just built a leaking bucket.
The one thing I don't get with showers is why there always is water left in them. I mean, I get the physics of it, I just don't understand why I can't just press a little lever on the shower battery and make the remaining water flow out in the drain. Then I can be sure that no water will slowly go out of my shower and make a damp bathroom, chalk stains or grow bacteria after I'm done showering.
I don't know if this is bash or an OS X feature, but nonetheless: today I discovered that there's tab-completion that takes .ssh/config into account! I on-instinct did a tab after writing part of the hostname I wanted and (boom!) there it was, auto-completed with a colon. So to SSH to my mini I did "ssh mi
In Scott Larsons blogpost referenced in my last post he gives a short but important note: his favourite MDX resource. And I agree: a 62 article series in learning MDX. Since I've worked with SQL "all my life" I've not bothered to much with MDX, but with this series I'm really looking forward to learning it better. To do the tutorials you'll need the free MS-SQL 2000 samples (that unfortunately aren't bundled with neither MS-SQL 2005 nor MS-SQL 2008). You can download them from here. I should note that while it compiles fine with VB6, using it with Visual Studio 2005 left me with a pretty large debugging job. But don't worry, the data sources are included and that really is all you need to work through the tutorials. For the interface, you've got BIDS
One of my main gripes with Microsoft SSIS is that there is no way to reuse logic. In my data integration task I needed to do the same lookup and translation tasks (typically convert to upper case, replace " with 'N/A', look up column in side table and use the IDs from that table instead) many times, in my case when importing data from an Axapta database. CozyRoc got back to me and told me that they have released an SSIS component that includes new components for reusability of code and flows. Being a coder I had given up on SSIS and rewritten my work in C#/SQL, but next time I'm very much looking forward to using it.
What bugs me very much about working with KPIs in PerformancePoint's Dashboard is that you cannot do simple calculations. For instance, I have a sales cube that has the measures unit cost and price the unit was sold for. I would like to say that a loss (price/cost < 1) makes the KPI red, a <20% margin makes it yellow and >=20% is green. In Excel this would be trivial: choose the price column, type '/' and choose the cost column: voila, you've got your calculation. Not so with Dashboard. As Rex Parker from Microsoft writes in the PerformancePoint Team Blog you'll have to write an MDX query. I don't believe MDX is for the average business person that PerformancePoint Dashboard is aimed at. But until MS puts together an expression tool that is as least as understandable as Excel, Scott Larson has put together a nice little tutorial to show how to write such expressions with MDX.
At work I'm looking at PerformancePoint, and being a newbie with this product I of course do all the newbie errors. That's why I was so happy when finding Nick Barclay's blogpost where he explains all about the way-to-common connection errors. Big thank you to Nick.
Not quite happy with any of the programs I found to identify iTunes duplicates, I spent an hour making my own (beats tracking down duplicates. ) I thought I'd share it with you. It's my first attempt at using OS X' ScriptingBridge and written in Python. I have no clue if it runs out of the box, I suppose you should have Developer Tools installed. It's not the fastest beast either, and Python and iTunes both use 50% CPU. But it gets the job done. If you wonder why I give so many parameters to the track class I should say that I plan on reusing it to do some more iTunes housekeeping. The script will mark all the duplicates with one star. Then I can round them up and delete them afterwards.
Good fun, I just found another blog titled Not_an_Artist
I'm a big fan of Amazon EC2 and S3. To be able to use the amount of computing power you need and the amount of storage you need and only pay for what you use is a good way of computing in my view. However, EC2 wasn't built to run clustered databases like MySQL on, and I've had problems finding any service that will let me pay for the capacity I need (cpu time and disk space) and give me a guaranteed query time. Amazon's SimpleDB seems to be a good step in that direction, and I'm really looking forward to toying with it and hopefully use it for many interesting projects.