The Journey to GSoC 2018 - just the beginning

Here I’ll document some of the preperatory tasks I undertook as part of my proposal to RTEMS (AN RTOS THAT’S BEEN TO SPACE! SPAAAAAAAAAACE!).

The Hunt

To start at the beginning, I’m going to have to just say that finding an organization that piques your interest and needs a project that hits just the sweet-spot in “I have most of the skills to do this, but I’ll still need to learn and grow a lot to bring it to fruition” is hard. I oscillated between organizations and potential projects for a month easily, until I finally stumbled upon RTEMS, at which point I yelled “I’m going to Mars!” repeatedly while jumping around the house. (Contributing to anything that’s even remotely related to space-research has been a dream for years, and the fact that RTEMS has such an incredible and approachable community has been incredibly exciting to me! :D)

Despite the excitement, RTEMS intimidated me majorly - their build system (using Autotools) is more complicated than any I’ve worked with before, and it all felt like magic that I definitely wasn’t cut out for:

Rise from the ashes

I decided eventually that I’d rather suppress my fear of looking stupid and just use the community’s knowledge whenever I needed it, by asking questions and learning instead of struggling in vain.

This resulted in very productive discussion on the mailing list, and even a few upstreamed patches to the core repository (the first 4 commits there), which let me finally gain some momentum.

Given that I was still working part-time (on this holy-shit-how-can-you-do-that project) and attending university, I decided I’d allot about 1 day per week to furthering my familiarity with RTEMS and figure out which specific project I’d like to work on.

This worked out surprisingly well! I started working towards my proposed phase 1 target of getting a new stub port for RTEMS (i.e. one that implements empty function definitions as much as possible, just to get it linking with the core of RTEMS, and therefore eventually all of the testsuite).

In the process of doing this, I came across a teeny tiny bug in 👏 G 👏 C 👏 C 👏. That’s right, I had the opportunity to submit a teeny patch to GCC for the x86_64-rtems-gcc target to allow it to include the standard RTEMS tools flags like -qnolinkcmds, -qrtems, etc. through the GCC spec syntax.

The patch that will let me die happily ever after.

After that, I continued working on the stub port and got some RTEMS tests linking to my stub happily, but there’s still tons of work to be done. See my post detailing my GSoC 2018 proposal for more information on what’s coming next.