June's first post

I am surprised to find out that I did not do much blogging in the month of May. At least I made a status report to Hassen and Jan. Time to write this chapter.

Time to be active on blogging and writing of industry experiment

It looks that the TDD project begins to slow down. There were less development activities in last 2-3 weeks in the TDD project. Interestingly, non-TDD is still active.

Summary of weekly telemetry report

Parameters:
Project: The-NON-TDD-Project
Week: 11-Mark-2007 to 17-Mar-2007 ~ 22-Apr-2007 to 28-Apr-2007
Report : TDDProject-Member-Report

The development slows down too in the month of April. It looks like the TLD developers tested somewhat less than the TDD group. It is a pity that the percentage of TDD is not up to the point.

Anyhow, it is interesting to see this difference.

Summary of weekly telemetry report

Parameters:
Project: The-TDD-Project
Week: 11-Mark-2007 to 17-Mar-2007 ~ 22-Apr-2007 to 28-Apr-2007
Report : TDDProject-Member-Report

Starting of April, the development slows down a bit observed from this chart. The effort spent on test is about a quarter of the effort spent on production code. It is interesting to know that developers never stopped working on tests. :-)

The length of episodes varies from 11 minutes to 41 minutes in this period. There is no pattern found yet.

The percents of TDD by time and number are both small. A thorough review of development method is necessary in order to tailor Zorro's rules for them, or they could do TDD better.

Test Last Project did not catch up tests yet

Test first project did not write enough test in the previous post, the test last project is even worse. Test must be forgotten by developers.

Production and Test DevTime

This picture is the telemetry stream of Total, Production, and Test development time of the test first project. Apparently the TestDevTime is significant lower than the time spent on production code in weeks 1, 3, and 4. More time was spent on test in the week 2.

Percentage of TDD is not good

Except for one developer, the rest all have 0% TDD percentage.

Status report and T/P ratio

I figured out that continuously reporting who has installed the sensors and who has not was not necessary after a month's pilot. So far, two developers never installed the sensors. Most developers who have installed the sensor are sending data to the server.

I configured the telemetry stream displaying the ratio of DevTime spent on the test code vs. DevTime spent on the production code. A value of one indicates the developer spent equal amount of time on test code and production code.

Range	Number
>1	2
0.5-1	2
0.2-0.5	2
0.1-0.2	5
<0.1	2

2007-04-11: 12 Developers have sent data to the server since 04/01

Though it is a bit disappointing, 12 is already 60% of participants. :-)

2007-04-10: Only one developer has TDD%

According to my data analysis, Zorro inferred that only one
developer conducted TDD development in 74.2%.

2007-04-10: Five users sent data to server today

Five users from Norway sent data to server. I double checked and found that the historical data was recovered.

2007-04-09: Only four users sent data to server

I discovered that server had OutOfMemory problem on April 5th. The Hackystat sensors can serialize sensor data locally when sever is unreachable. So, technically, there should have no data loss. According to today's report, only four users have sent data to server today. Two of them are from Norway, and the other two are from Vietnam. Heap size of the server is about 180MB.

2007-04-05: Server restart notice

Below is a copy of the email I sent on April 5, 2007. Some sensors data between 03/31 and 04-05 were missing.

Hi, All,

I just noticed that the Hackystat server had a memory error. I restarted the Tomcat.
It works now.

Thanks,
Hongbing

En route to the industrial case study description

From now on, this will be the place where I put the Norway case study materials.