Friday, August 20, 2004

Week 8 - Back at the ranch

This week, we're slummin' at the home office.
Doing more code investigation and prepping for the project kick-off.

Looks like we have to dramatically cut scope, since what looked like a 3 month project likely will take more than a year. Incidentally, the client seemed not at all surprised !

We can automate a lot of the gruntwork, but that sort of defies the secondary aim of the project, which is teaching the client how to start developing "agily".

Reading Jim Highsmith's book Agile "Project Management"
(see http://www.jimhighsmith.com/pubs.html)
Salient detail #1: my colleague Robin Gibson collaborated on that, you know.
Salient detail #2: our CEO, Roy Singham, wrote a comment that's printed in and don the book.

And who joins us for lunch at the office ?
By Toutatis, it's mr Highsmith himself !

I better get to finishing the book before Monday...

Monday, August 09, 2004

Week 7 - Finishing the preparations

The client has a varied codebase, with servlets and JSPs.

Some of these are using struts.

What are struts ?

http://struts.apache.org/

Apparently, this is a way to abstract model from view better than ordinary JSPs do.

A note about a cool MS Excel trick:

If you want to chart a bunch of data points that do not form a smooth line, don't knock yourself out trying to deduce the 3rd order polynomial !

Select the data range and make an XY scatter chart, maybe even select the subtype with smoothed lines. Now select one of the datapoints in the chart and right-click.

Choose "Add Trendline..." and select one of the types that looks closest to your data.

Select the Options tab and switch on "Display equation on chart".

Now you have your polynomial and can eat it too. Select the text of the equation and paste it into the cells where you want to get missing data. Modify it to include your source cells and there you go.



Monday, August 02, 2004

Week 6 - The knowledge grows

My colleagues are launching new stuff at me at such velocity, I can hardly keep up.
Wow, I though I knew a thing or two about software, but I guess I've been asleep for the last two years.

A few notes on some very useful tools to better estimate project duration.

Learn about Wideband Delphi
http://www.processimpact.com/articles/delphi.html

Tom DeMarco and Timothy Lister wrote a book about project management from a
risk management perspective, "Waltzing with bears".
http://www.dorsethouse.com/books/waltz.html

If you want to jump into the math, look here:
http://mathworld.wolfram.com/RayleighDistribution.html

Test code coverage can be done using Clover or JCoverage or EMMA.
Here's a user's testimonial on all three.