| 22 Dec 2025 |
|  @cinerealkiara:matrix.org left the room. | 10:53:47 |
| 23 Dec 2025 |
 piegames | Hm. Unfortunately, the Nix language is not amenable to single-parse-pass compilation because of parsetime attrset merging | 09:56:00 |
| piegames | Everything else can be done with placeholders and a fixup pass afterwards, but attrsets really require using some treemaps during construction | 09:56:40 |
| piegames | Also let's move this over to Zulip https://zulip.lix.systems/#narrow/channel/6-Eval/topic/Parser.20rewrite/with/3389 | 10:36:24 |
 XYenon | Should I do anything now, or just wait? My CL hasn't had any new reviews for a week.
https://gerrit.lix.systems/c/lix/+/4783 | 13:44:10 |
| piegames | Asking here is fine | 14:10:07 |
| piegames | Oh I see, there has been a disagreement between reviewers. Maybe we should escalate it to a core team discusssion then | 14:11:59 |
 raitobezarius | where? | 14:15:48 |
| piegames | The CL has both +1 and -1 reviews | 14:18:02 |
 helle (just a stray cat girl) | is that not just purely for the test impl | 14:18:14 |
| raitobezarius | i think you are misreading the reviews | 14:18:13 |
| piegames | Or am I reading it wrongly (I'm on the phone so lol@Gerrit UI) | 14:18:19 |
| raitobezarius | +1 was given over the F2 part | 14:18:40 |
| raitobezarius | -1 was given over the actual contents | 14:18:46 |
| raitobezarius | * -1 was given over the non testing changes | 14:18:49 |
| raitobezarius | anyway, this stalled because:
(a) flakes are frozen in principle and we recommend reaching out here before sending changes
(b) selected reviewers did not interact with the CL
(c) it's holidays for many people :P | 14:19:59 |
| XYenon | Get it. I thought this was just a small backport change. | 14:47:52 |
| raitobezarius | No worries, unfortunately, as we touched upon it in https://lix.systems/blog/2025-11-18-lix-2.94-release/ (section "Flakes freeze"), too many inconsistences kept being discovered in how these features are rolled in CppNix | 14:49:37 |
| raitobezarius | And as you have seen, horrors flagged such a problem on that backport, alas | 14:49:57 |
 K900 | Oh yeah does anyone want a trivial UX improvement | 15:16:00 |
| K900 | Make it so -L doesn't print the failing drv logs twice | 15:16:24 |
| K900 | For bonus points make it so it prints the full list of all failing drvs with keep-going | 15:16:42 |
| raitobezarius | open an issue so that someone gets to it eventually | 15:19:25 |
| K900 | I might just do it myself if no one else does | 15:19:39 |
| K900 | Not tonight though | 15:19:49 |
 ma27 | fwiw it should be as simple as setting log-lines=0 if -L is set (iirc I even suggested that to cppnix a while ago, but eelco said no).
I have this in all of my nix.conf and I think that's enough to do what you want:
error: build of '/nix/store/bc0s2qws1gnhqqn68dwpm5ypafkichwm-vim-9.1.1869.drv' on 'ssh-ng://roflmayr' failed: builder for '/nix/store/bc0s2qws1gnhqqn68dwpm5ypafkichwm-vim-9.1.1869.drv' failed with exit code 1
error: builder for '/nix/store/d0xjfxgyj5h2girabscd1ss8bsmx9b2v-hello-2.12.2.drv' failed on remote builder
error: builder for '/nix/store/bc0s2qws1gnhqqn68dwpm5ypafkichwm-vim-9.1.1869.drv' failed on remote builder
error: build of '/nix/store/bc0s2qws1gnhqqn68dwpm5ypafkichwm-vim-9.1.1869.drv^out,xxd', '/nix/store/d0xjfxgyj5h2girabscd1ss8bsmx9b2v-hello-2.12.2.drv^out' failed
| 15:27:58 |
| K900 | Yeah but then you don't get the logs | 15:28:18 |
| K900 | Or what | 15:28:23 |
| K900 | Wait am I dumb | 15:28:33 |
| K900 | Wait does log-lines | 15:28:36 |
