berdario/gist:09a59cf0f2c2a51c168225cca5f1dd0d Secret

## gistfile1.txt

The Oceansprint was great: My last contribution to nixpkgs was unfortunately back in 2018, so joining this event has been a great way to get back up to speed on the project, the Nix language itself and the problems that people are trying to solve right now.

I started the week by looking into some Darwin build failures (for cloc and the agg library), I then updated my home-manager configuration, used nixpkgs-review for the first time... and finally upon Bernardo's suggestion I started to look into factoring out common attributes used in kernel modules derivations. This would help people use the correct flags to build them (especially needed when changing the compiler used to build the kernel). I started by simply merging the attrs, and then I drafted a more involved PR. After trying to apply this change to ~60 different modules, I realized that we have too many different attrs, which are also often going to be arbitrary bits of shell used in the different build phases, and thus trying to restrict them won't work well. To get the PR ready I'll thus move back to the simpler initial approach.

The event went smoothly, despite all of the Covid uncertainty. I especially appreciated how everyone tested themselves at least twice during the duration of the sprint, giving everyone peace of mind that the social interactions wouldn't be excessively risky (and anyhow everyone was outdoor most of the time).

	The Oceansprint was great: My last contribution to nixpkgs was unfortunately back in 2018, so joining this event has been a great way to get back up to speed on the project, the Nix language itself and the problems that people are trying to solve right now.

	I started the week by looking into some Darwin build failures (for cloc and the agg library), I then updated my home-manager configuration, used nixpkgs-review for the first time... and finally upon Bernardo's suggestion I started to look into factoring out common attributes used in kernel modules derivations. This would help people use the correct flags to build them (especially needed when changing the compiler used to build the kernel). I started by simply merging the attrs, and then I drafted a more involved PR. After trying to apply this change to ~60 different modules, I realized that we have too many different attrs, which are also often going to be arbitrary bits of shell used in the different build phases, and thus trying to restrict them won't work well. To get the PR ready I'll thus move back to the simpler initial approach.

	The event went smoothly, despite all of the Covid uncertainty. I especially appreciated how everyone tested themselves at least twice during the duration of the sprint, giving everyone peace of mind that the social interactions wouldn't be excessively risky (and anyhow everyone was outdoor most of the time).