| 16 Dec 2022 |
|  Growpotkin joined the room. | 16:54:47 |
| Growpotkin | Robert Hensing (roberth): beat me to the punch on just about every comment I was going to add to the RFC haha. Thanks for saving me the effort. | 16:59:56 |
| Growpotkin | The only area that I would expand ( possibly in a footnote ) is "why pkg-fun.nix instead of default.nix?".
On the surface this change likely seems the largest to an average reader.
Something along the lines of "default.nix was never really the appropriate filename, this change leaves default.nix available for it's original purpose for packages which choose to use it, etc..." | 17:02:23 |
| Growpotkin | * The only area that I would expand ( possibly in a footnote ) is "why pkg-fun.nix instead of default.nix?".
On the surface this change likely seems the largest to an average reader.
Something along the lines of "default.nix was never really the appropriate filename based on Nixpkgs' usage. This change leaves default.nix available for it's original purpose for packages which choose to use it, etc..." | 17:02:59 |
| Growpotkin | There's likely a way to phrase it that avoids philosophical debate about what default.nix ought to be used for though. It is the internet after all so I'd hate to open up a can of worms that derails relevant discussion. | 17:05:04 |
| Growpotkin | Robert Hensing (roberth): any preference on the "common" helper routine for readDir and readFileType living in primops.cc vs util.cc? ( basically DT_* -> char * part ) | 19:45:34 |
| Growpotkin | * Robert Hensing (roberth): any preference on the "common" helper routine for readDir and readFileType living in primops.cc vs util.cc? ( basically DT_XXX -> char * part ) | 19:46:41 |
 Robert Hensing (roberth) | Growpotkin: I think the char * part is specific to the language, so util doesn't seem right | 19:48:41 |
 growpotkin1 | Sounds good. | 19:48:54 |
| growpotkin1 | Related : should we be using a symbol here to avoid allocating a billion strings? | 19:49:29 |
| Robert Hensing (roberth) | Yeah, good idea | 19:50:02 |
| growpotkin1 | Like the ones stashed in baseEnv | 19:50:14 |
| Robert Hensing (roberth) | Oh, adding constants to the language? Not sure about that | 19:50:56 |
| Robert Hensing (roberth) | That's not something we have generally done | 19:51:15 |
| Robert Hensing (roberth) | Yeah not a single string constant there it seems, unless you count the nix version | 19:52:37 |
| Growpotkin | we have symbols for null, true, false, etc. That was what I meant ( EvalState has symbols for sWith, sOutPath, etc ) | 19:58:54 |
| Growpotkin | * we have symbols for null, true, false, etc ( constants ). But we also have symbols in state - that was what I meant ( EvalState has symbols for sWith, sOutPath, etc ) | 19:59:22 |
 infinisil | Wait there's two growpotkins?? | 20:22:47 |
| Growpotkin | yeah I'm a dingus and logged in to some devices with google and others with a user/password and I have no idea how to merge the two matrix accounts it created | 20:23:25 |
| Growpotkin | This is the account that my personal computer has. The other one is my work computer and phone. | 20:23:55 |
| Growpotkin | https://github.com/NixOS/nix/pull/7447/files Robert Hensing (roberth) look upon my works ye mighty and despair | 20:24:32 |
| Growpotkin | * https://github.com/NixOS/nix/pull/7447/files Robert Hensing (roberth) look upon my works ye mighty and despair ( let me know if you have more notes lol ) | 20:24:44 |
| Robert Hensing (roberth) | ah mkString doesn't copy, so no need for symbols | 20:28:10 |
| Growpotkin | If the function name is a real problem for ( someday ) merging lazy trees let me know and I can rename it. Personally I like that the name indicates that a filesystem read will occur, and that it can't be confused with "constructing a pathType primitive"; but it's not a deal breaker for me. | 20:29:55 |
 raitobezarius | Hey there, is https://github.com/NixOS/nixpkgs/issues/206467 in the scope for the architecture team? | 23:25:07 |
| 19 Dec 2022 |
| Robert Hensing (roberth) | raitobezarius: we focus mostly on the pkgs side of things, but maybe someone is interested? | 14:49:55 |
| Robert Hensing (roberth) | (I have too much on my plate rn, but I would recommend to inventorize the current implementations, the problems they have, what it takes to solve them and how NixOS can support the stateful operations by taking care of the state part) | 14:51:56 |
| Robert Hensing (roberth) | (because, just like terraform needs it, NixOS will need to manage its own state to keep track of what it manages and what it doesn't) | 14:52:33 |
 John Ericson | Sorry I'm travelling cannot make it today | 15:07:19 |
| infinisil | @room: The next meeting will take place in about 6 minutes, we'll continue discussing https://github.com/nixpkgs-architecture/simple-package-paths - meeting link - live stream - meeting notes | 15:24:24 |
