Starbucks-fueled Developer

Monday, May 29, 2006

6 Seconds of Fame

As may[not] have read previously, my brother is currently living/studying at Full Sail in Central Florida to become a computer animator. He'll be starting his third month tomorrow and the fruit of his 3D Foundations is available for viewing on his blog (or directly downloadable; again, apologies for the URL...). Watch and return...

Some spec's not mentioned in his post: he mentioned that it's only 6 seconds long. At 30 fps, that's 180 frames...which took nearly 3 hours to render/compile/encode using QuickTime and produced the 18.5 MB that you (hopefully) viewed. That boggles my mind. If you take a feature-length film, like Incredibles - that runs nearly 2 hours - you're talking about 216,000 frames.

Remember: his doesn't have textures either. His rendered frames per hour (Rfph) is 60. Now, granted, Disney is going to have supercomputers working to render their films. But on the G4s that he's using*, that's 3,600 Rfph, or 150 DAYS!!!

They desperately need to refactor their Continous Integration process... ;-)


*sorry, I had to get this in...Sean mentioned that his PC (an older Dell P4 3.0GHz HyperThreading Wintel) is faster than the Macs for this stuff...OK, I'm better now...

Thursday, May 25, 2006

Process Problems

While this won't be an issue for me much longer, I recently ran into a situation that I'm not sure how agile-ists handle during product/project development. I'll try to summarize what happened.

A project I was working is heading into its "maintenance" phase; essentially we completed Release 1 and are continuing to support the application as bugs and new features are found and requested by the customer. We sat down with the customer to prioritize what work should be done from their backlog for the first post-production iteration. At the time of meeting, there were still several (read, over 15) remaining outstanding/undelivered/unresolved/un-developed bugs that had not been tackled by the team. Now, my impression is that these bugs would be in the top of the stack/queue for work during the start of the coming iteration. However, the problem arose that it was the opinion - and rightfully so - that bugs should be addressed immediately once discovered.

Here's what I don't understand: the bugs will require us to investigate/triage/whatever to determine how much effort it will take (hours, days, etc.) to correct the problem. Obviously, the investigation is a non-trivial effort (read, legacy code). Given this information, where/when during our iteration(s) do we actually investigate the new bugs as the time to investigate is taken away from the development time we allotted for fixing bugs and/or implementing new features?

My initial thought was, "we'll have to do the estimates for the new ones at the end of the current iteration as part of our retrospective"; obviously that doesn't necessarily work because if we're on Day 2 of a 30-Day iteration, that leaves 28 or days before the customer sees any visible progress on the bug.

Not good.

So am I missing something here and making it too complicated? Obviously, the customer has stated that they want bugs fixed over new features. But it does feel that simple...or is it??

Wednesday, May 24, 2006

IntelliSense for SQL Server Tools

I went for a visit to Scott Guthrie's blog to get the Web Application Project Template and found that Red-Gate Software now has a free (like Visual Studio Express free) tool for providing IntelliSense in the vairous tools for SQL Server. See Red-Gate's site for more information.

While I haven't been writing much SQL recently, this is a nice tool to have...especially since it's free...

Tuesday, May 16, 2006

Goodbye GotDotNet, Hello CodePlex

Rob Caron announced, along with James Newkirk and Korby Parnell, that Microsoft has deployed the first public beta of the replacement for GotDotNet.com - CodePlex.com, which is build on Team Foundation Server.

It appears that membership is free, but I was not able to find any information on whether or not members must purchase a license of Team Explorer to connect to CodePlex or how that'll work out. However, since the Express Editions do not support MSSCCI out-of-box, it looks like dev's will have to pony-up for Standard Edition to integrate the IDE with CodePlex using the Team Foundation Server MSSCCI Provider....at least that's my current understanding...

Should be interesting to see how this develops...