| 24 Mar 2026 |
 delroth | it's an stdlib problem more than a C++ problem, most large C++ codebases at the time were using a Result type in 2005, it's just that that knowledge wasn't widespread yet or too publicly available through libraries | 10:25:06 |
 piegames | Granted, but still sum types in C++ are a lot less ergonomic than they could be | 10:26:04 |
| delroth | yes, but you don't need full sum types capability for a Result type, you basically need the "?" operator | 10:26:56 |
| delroth | * yes, but you don't need full sum types capability for a Result type, you basically need the "?" operator / syntax sugar | 10:27:22 |
| * delroth has written way too much code with ASSIGN_VALUE_OR_RETURN and util::StatusOr which dates back to ~2003... | 10:28:31 |
| * delroth * has written way too much code with ASSIGN_OR_RETURN and util::StatusOr which dates back to ~2003... | 10:28:56 |
|  Lotte (it/its)/Cinny (she/her) θΔ& changed their profile picture. | 12:15:06 |
 ShalokShalom | Why C++ to implement a purely functional language at any point? | 16:01:45 |
| piegames | What does the one have to do with the other | 16:17:19 |
| 25 Mar 2026 |
 Lisanna | Is it just me or does Lix compile faster than ccpnix? | 16:41:17 |
| Lisanna | * Is it just me or does Lix compile faster than cppnix? | 16:41:28 |
 Qyriad | I haven't compiled cppnix in a while, but I'm not all that surprised. jade in particular put some effort at the start in reducing compile times, and cppnix also made their meson buildsystem more complicated because they wanted to separate it into multiple derivations | 16:51:29 |
| Lisanna | yeah I am really sad about the multiple derivation thing, idk what value it really adds and it makes overriding things hell | 16:56:34 |
| Lisanna | (see: https://gist.github.com/lisanna-dettwyler/ba37119993618314891d5afd79a8cf5e) | 16:56:50 |
| Qyriad | In reply to @lisanna-dettwyler:matrix.org
yeah I am really sad about the multiple derivation thing, idk what value it really adds and it makes overriding things hell the motivation is to be able to re-run derivation tests, like the NixOS VM tests, with faster compile times if you only changed one library. theoretically | 17:03:19 |
 Sergei Zimmerman (xokdvium) | In reply to @lisanna-dettwyler:matrix.org
Is it just me or does Lix compile faster than cppnix? FWIW I picked jade’s awesome changes. Make sure to use clang | 17:16:46 |
| 26 Mar 2026 |
 aloisw | raitobezariusdo you know for sure now that the SQLite changes you linked in the Zulip thread make things better? | 17:42:31 |
 raitobezarius | i'm really sorry, i don't know yet | 17:43:10 |
| raitobezarius | it's somewhere on my todolist to apply to the CI of AFNix globally | 17:43:19 |
| aloisw | No problem, your statement just sounded rather authoritative to me. | 17:45:09 |
| raitobezarius | you're right, i should have made expectations clearer this is untested | 17:46:14 |
| raitobezarius | but also, i wanted to communicate there are tools we did not fully explore yet that can act on SQLite | 17:46:27 |
 Winter | i love cppnix and lix fighting over flake lockfiles:
diff --git a/nix/flake.lock b/nix/flake.lock
index 1da808bbd1..dd4c77ea33 100644
--- a/nix/flake.lock
+++ b/nix/flake.lock
@@ -44,13 +44,13 @@
"rev": "c2520574da83f8e7de1d781ec64d8722c75a515d",
"revCount": 427,
"type": "git",
- "url": "https://github.com/hegeldev/hegel-core?dir=nix"
+ "url": "https://github.com/hegeldev/hegel-core"
},
"original": {
"dir": "nix",
"ref": "refs/tags/v0.2.3",
"type": "git",
- "url": "https://github.com/hegeldev/hegel-core?dir=nix"
+ "url": "https://github.com/hegeldev/hegel-core"
}
},
"nixpkgs": {
i imagine this is cppnix's fault as the ?dir=... only gets added in newer versions (>2.28 or so) and lix keeps trying to delete it, probably to maintain proper compatibility, but wow this is annoying.
@Sergei Zimmerman (xokdvium) is this a valid bug report for y'all or am i going to tear my hair out ad infinitum?
(posting here instead of #Nix Package Manager development bc idk it feels like the better of the two options lol) | 18:16:34 |
| Winter | i love cppnix and lix fighting over flake lockfiles:
diff --git a/nix/flake.lock b/nix/flake.lock
index 1da808bbd1..dd4c77ea33 100644
--- a/nix/flake.lock
+++ b/nix/flake.lock
@@ -44,13 +44,13 @@
"rev": "c2520574da83f8e7de1d781ec64d8722c75a515d",
"revCount": 427,
"type": "git",
- "url": "https://github.com/hegeldev/hegel-core?dir=nix"
+ "url": "https://github.com/hegeldev/hegel-core"
},
"original": {
"dir": "nix",
"ref": "refs/tags/v0.2.3",
"type": "git",
- "url": "https://github.com/hegeldev/hegel-core?dir=nix"
+ "url": "https://github.com/hegeldev/hegel-core"
}
},
"nixpkgs": {
i imagine this is cppnix's fault as the ?dir=... only gets added in newer versions (>2.28 or so) and lix keeps trying to delete it (while older cppnix's fetchgit just cries and throws up if it sees it), probably to maintain proper compatibility, but wow this is annoying.
@Sergei Zimmerman (xokdvium) is this a valid bug report for y'all or am i going to tear my hair out ad infinitum?
(posting here instead of #Nix Package Manager development bc idk it feels like the better of the two options lol) | 18:17:27 |
| Sergei Zimmerman (xokdvium) | Hm I seem to recall https://github.com/NixOS/nix/blob/master/src/libflake/flakeref.cc#L304C5-L346C21 was trying to fix a similar issue? | 18:20:35 |
| Winter | A nix flake lock by... 2.22.1-- wait, why the fuck is this using such an old version? | 18:23:19 |
| Winter | christ | 18:23:23 |
| Winter | okay clearly I have slightly bigger issues to worry about | 18:23:29 |
| Winter | thanks for the code pointer and sorry for the noise 🙃 | 18:23:54 |
| Winter | will come back if there's something substantial when i'm not using an ancient nix | 18:24:08 |
