2023-11-27 Nix team meeting minutes #107

fricklerhandwerk · November 27, 2023, 4:08pm

Agenda

@fricklerhandwerk didn’t have time to finish the statement of work
- got the ball rolling on general teams funding though, will discuss it at the next all-teams meeting
@tomberek: do we need to make a formal request to the board?
- @thufschmitt: currently no plans for direct funding by the foundation
  - also no capacity or intention to do any sort of micro management
  - details for any proposal need to be laid out publicly
@tomberek will open an issue on the foundation’s tracker
(some debate on making teams more accountable)
@tomberek: propose to start with giving formal teams $500, people will self-organise around the responsibility

@Ericson2314 wrote up a draft
@fricklerhandwerk: doing an in-depth review will likely stall progress because we will surely find controversial detail whem looking closely enough
@fricklerhandwerk: proposal for a collective statement:
- the team likes the general direction, and would like to see community feedback and give the proposal time to mature
- it’s important but not urgent
  - in particular, less important and urgent than (in order of increasing importance and urgency)
    - closing the flakes case
    - stabilising the new CLI
    - bugfixing and finishing things we started
@thufschmitt: eventually this is RFC-level material
- @fricklerhandwerk: recommend starting with a Discourse post and, maybe afterwards, a separate RFC repo without opening an RFC pull request, so there is a space to collect inputs and discuss
no disagreement on these ideas

Switch from std::regex to boost::regex by yorickvP · Pull Request #7762 · NixOS/nix · GitHub makes us depend on a custom build of boost that may not be in the cache
- anyone who wants to build Nix will have to build boost
- the overridden variant is not in Nixpkgs: Split boost into multiple outputs · Issue #45462 · NixOS/nixpkgs · GitHub
- @edolstra: since it’s part of the Nix build closure, which is built by Hydra, it will also end up in the cache
@thufschmitt: then it’s a non-issue that will disappear in a few days

@Ericson2314: need to finish moving things around for documentation rendering
@roberth: no progress on the structured arguments. should probably do the code churn first
@fricklerhandwerk: no tangible progress
- briefly discussed the design issues with @infinisil
- briefly talked to @nmattia (niv author)
- have to consider cost of change in general
- will do some prototyping to get a better feeling for the problems

@kolloch’s proposal for the behavior is sensible
@fricklerhandwerk: we may want to say algo, because we’ve run into confusing type with encoding
- @edolstra: prefer algo
- @Ericson2314: we call internal variables algo but no one bothered to change the field names