| 13 Oct 2025 |
 Yureka (she/her) | How? | 19:40:47 |
 hexa | narinfo-cache-positive-ttl = 3600;
| 19:40:59 |
| Yureka (she/her) | on the client? | 19:41:13 |
| hexa | yeah | 19:41:16 |
| hexa | Redacted or Malformed Event | 19:41:20 |
| hexa | * with gcing caches you should set a shorter positive lifetime for the cache | 19:41:44 |
| Yureka (she/her) | yes yes I know the option, but I don't like that there is no way to fix this from the server (like remotely advertising a ttl) | 19:41:58 |
 raitobezarius | either way, lix shouldn't crash on actually 404s NARs | 19:42:18 |
| Yureka (she/her) | but even with a low TTL, it's not "safe" to gc | 19:42:22 |
| hexa | yep | 19:42:30 |
| raitobezarius | it should gracefully consider that nar nonexistent | 19:42:31 |
| raitobezarius | then you can also evict it from the narinfo cache as well | 19:42:42 |
| raitobezarius | this should be at most inefficient | 19:42:49 |
| Yureka (she/her) | I thought I fucked up the integrity of my cache, so I entirely recreated the cache and re-populated it with contents from the old cache, but it kept crashing | 19:42:58 |
| Yureka (she/her) | so now I know this is just normal(TM) | 19:43:17 |
| Yureka (she/her) | after some hours of debugging | 19:43:19 |
| raitobezarius | i think we don't need to keep up the act and we can say it out loud | 19:43:21 |
| raitobezarius | any cache | 19:43:22 |
| raitobezarius | in | 19:43:23 |
| raitobezarius | this codebase | 19:43:26 |
| raitobezarius | should be considered your enemy | 19:43:30 |
| raitobezarius | (this is almost not even a joke, i cannot count how many unsound caches we have in the codebase, some known, some we stumble on like "wait what?") | 19:43:54 |
| Yureka (she/her) | enemy nr. 1 - the eval cache
enemy nr. 2 - the narinfo cache
enemy nr. 3 - cache.nixos.org | 19:43:55 |
 Marie | with my recent hydra shenanigans, i agree | 19:43:59 |
| raitobezarius | enemy nr. 4 — https://gerrit.lix.systems/c/lix/+/4071/3 | 19:44:11 |
| raitobezarius | enemy nr. 5 — /nix/var/nix/db/db.sqlite | 19:44:29 |
| raitobezarius | (ok 4 & 5 are almost the same) | 19:44:49 |
 ma27 | what did I miss?
anyways, yes I did, but my patch actually caches less in it (because the NAR url isn't necessarily what lix expects, e.g. with snix-castorè).
but if this is something I should take a look at, I'll have to defer to tomorrow anyways. | 19:55:41 |
| raitobezarius | np | 19:56:14 |
 KFears (burnt out) | I wonder if Nix channels can be considered number 6 (I have today stumbled upon the cursed immutable tarballs protocol and I'm very unhappy) | 19:57:38 |
