Scilab

Scilab is an open source software for numerical computation. What this means is that it provides you with all the mathematics formulaes you need to modelize whatever you need. An example of this would be to modelize how a sheet of metal will bend under pressure, when applied here or there, and see the distortion and the most stressed locations. You know which formulae gives you the distortion, Scilab calculates how it is applied on the whole surface. This is a very simple example to give an idea, but it is also used in satellites testing, sound propagation analysis and pretty much anything where you can find mathematics and modelization.

When I joined Scilab, it was a project of a national research center. It grew to finally start a company of its own, Scilab Enterprises, which became the official editor of the software. I applied there for a position of System Administrator, which I had no idea how to handle, but I knew that I wanted to try and work for open source, it was as good a start as I was going to get, it was well paid and I had practiced a bit Scilab during my engineering school.

My responsabilities were:

  • keep the website up to date. On one occasion during the three years I spent there, I also had to rework the front-end to make it look shinier/2.0 era.
  • manage the machines and servers, including mail servers, web servers, build machines
  • take care of the production of nightly builds. These were produced from a custom built Perl script, modularized and optionized by myself to be able to get a version from scratch in less than an hour, for any of the 5 OS (Windows 32/64 bits, linux 32/64 bits, Mac OS).
  • take care of building the toolboxes (ATOMS modules) developped by the community and have them accessible from the ATOMS portal.
  • random IT related things, between bash development, css/html/php frontend development, automation, support to the community and test.

From where I stand now, I can't count how many things I would have done differently. At least bootstraping the websites (or anything, basically, that would allow responsiveness instead of what we had), progress from the perl scripts to real continuous integration, a more API driven approach for the toolboxes usage, just so many things had (or still have) room for improvement.

But even with these lacks, I've learn much and more about IT, development (web and soft and scripts), debugging, administration, that I cannot regret too much either. I did as best as I could, with all the knowledge I had or could find at the time.