[Nix-dev] how does NixOS ensure focus?

Vladimír Čunát vcunat at gmail.com
Mon Jul 28 11:07:46 CEST 2014


On 07/28/2014 03:55 AM, hasufell wrote:
> However, since I'v seen a lot of things gone wrong in distros (primarily
> gentoo), I am curious if you have a concept for ensuring that NixOS
> stays focussed.
> [...]
> A lot of projects (some successfully) ensure focus by limiting the
> number of core developers to the absolute minimum, which means
> decision-making is pretty easy. At the same time they open up the gates
> for random collaboration. This implies a well designed review workflow
> with very responsive and active developers for this to function
> properly, otherwise people will give up on overcomplicated contribution
> procedures/channels and do their own thing.
> I'd pretty much say that the linux kernel runs this model successfully,
> although reasons for this are probably a bit more complicated.

As has been said, there are no hard rules. We still just rely on 
judgement of those with commit rights and some agreements from before.

My view: the kernel model is infeasible for us (ATM). People and even 
companies use nix(os) to do what they want/need, and their improvements 
are contributed back (partly?). It's more of a symbiosis of people where 
each has his/her own "focus" and aims.

Decisions in such a model can become difficult, but fortunately nix(os) 
is very handy in supporting multiple different solutions independently.


Now a few clarifications:

On 07/28/2014 05:37 AM, Mathnerd314 wrote:
> Perhaps this has already started in the form of Guix; they have their
> own instance of Hydra, their own Guix packages, etc., but still use Nix,
> the underlying software, basically out of the box.

Guix doesn't use the nix *language*, and so they have their own 
counterpart of nixpkgs. IIRC they write expressions in guile, but 
generate the same primitive *.drv files, so they share the nix store and 
the process of building outputs from *.drv derivations. (i.e. the amount 
of sharing is relatively *very* small)

Personally, I consider it the only known real "fork" of nixos.org stuff 
(perhaps not in typical sense of fork).

> NixOS is pretty small now (only 21 people, according to
> https://github.com/orgs/NixOS/people) so there hasn't been much need to
> write it down.

True, the number of people with lots of contributions aren't high. In 
the github NixOS organization there are 36 people, actually, as some 
have selected "private membership". (I've got no idea what it's good 
for, when everyone can see what people commit, but I guess most people 
didn't notice/care about the choice.)


Vlada


-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 3251 bytes
Desc: S/MIME Cryptographic Signature
Url : http://lists.science.uu.nl/pipermail/nix-dev/attachments/20140728/f4094f2f/attachment.bin 


More information about the nix-dev mailing list