Thibauld - Imagination and Execution -

Freelance Business Club v1.0 released !

That's it! After 3 months of hard work, the new Freelance Business Club website has been launched this weekend... mission accomplished If you compare with the initial planning, we're 1 little week late... but, given the result, I think our last week decision to postpone the technical release was the good one!

For those of you who don't know what is Freelance Business Club, it is the first social network dedicated to IT freelancers (and their clients). It's a new concept which aims at easing the recruitment process of IT freelancers, it has been pioneered by Arnaud Coulondre (see his fbc profile) and Florent Daversin and I personally came onboard a bit later, after the launch of their website's beta version (you can see my own profile here).

My role in FBC, was to turn their beta version into a full-featured 1.0 version, which, I think, we achieved. Objectives were to lay the foundations for :

• Stable, predictable and maintainable software architecture
• Scalable technical architecture
• SEO efficient website (with the precious help of our beloved SEO master Sacha Pavan, creator of BankExam)
• Easy internationalization
• Integration with 3rd party webservices (Scribd for online CV visualization in our case)
• Easy integration of the new graphical chart (which I find awesome and was designed by Léo Robin)
• Implementing a performant, business focused and highly relevant search engine for FBC.

I'll probably use this blog to develop further what I did for FBC regarding each of the above points as I think it might be of interest.

Just for fun, I took some time to compare version beta and version 1 of the website and here are the results using SLOCCount:

Stats for version beta

Totals grouped by language (dominant language first):
php: 13258 (96.75%)
sh: 446 (3.25%)
Total Physical Source Lines of Code (SLOC) = 13,704
Development Effort Estimate, Person-Years (Person-Months) = 3.12 (37.49)
(Basic COCOMO model, Person-Months = 2.4 * (KSLOC**1.05))
Schedule Estimate, Years (Months) = 0.83 (9.91)
(Basic COCOMO model, Months = 2.5 * (person-months**0.38))
Estimated Average Number of Developers (Effort/Schedule) = 3.78
Total Estimated Cost to Develop = $ 422,018
(average salary = $56,286/year, overhead = 2.40).


Stats for version 1.0


Totals grouped by language (dominant language first):
php: 7556 (99.84%)
sh: 12 (0.16%)
Total Physical Source Lines of Code (SLOC) = 7,568
Development Effort Estimate, Person-Years (Person-Months) = 1.67 (20.10)
(Basic COCOMO model, Person-Months = 2.4 * (KSLOC**1.05))
Schedule Estimate, Years (Months) = 0.65 (7.82)
(Basic COCOMO model, Months = 2.5 * (person-months**0.38))
Estimated Average Number of Developers (Effort/Schedule) = 2.57
Total Estimated Cost to Develop = $ 226,241
(average salary = $56,286/year, overhead = 2.40).

I was impressed (not to say proud ) of the 45% decrease of the number of lines of code although version v1.0 has more features than version beta. Too bad SLOCCount doesn't take HTML or JS into account as I think results would have been even better... What's interesting also is that, given the number of lines of code found, sloccount uses the COCOMO model (a software cost estimation model) to "reverse estimate" the project main figures. For v1.0, according to the model, we should have spent 7.82 months with an effort of 20.10 person-month (that is 2,57 developers), to finish v1.0.

In reality, it took us almost exactly 3 months (we began on August 1st) to release v1.0 with 3 resources (among which 1 was not full-time, so let's say 2.57 resources to ease the comparison). Of course, comparison does not stand as v1.0 was not built from scratch and there was code (mainly the business layer) that was already developed for vBeta. But given the major overhaul of the code we did, I still think that releasing v1.0 was a very honorable performance!

But enough analysis already! Let's now have some beers and drink to FBC v1.0 before the first bugs appear

Santé !