| 29 Jun 2026 |
Grimmauld (any/all) | Consider: If hydra GC eats a dependency of a python package, and doCheck happens to be false, it'll break WAY WAY later if something is imported but required dependencies are lost as requisites | 16:51:09 |
K900 | Nothing to do with Hydra | 16:51:21 |
K900 | It's just failing tests in increasingly stupid ways | 16:51:31 |
Grimmauld (any/all) | the GC shenanigans worry me a lot with optional dependencies, but on python it could be even worse | 16:51:36 |
K900 | My money is on filelock | 16:52:44 |
K900 | Which they abuse in absolutely abhorrent ways | 16:52:51 |
K900 | But the fun part is that if I downgrade filelock, setuptools explodes | 16:54:51 |
Grimmauld (any/all) | python 3.14? | 16:56:04 |
K900 | Nope | 16:56:08 |
K900 | Same thing on 3.13 | 16:56:11 |
Grimmauld (any/all) | huh! | 16:56:16 |
Grimmauld (any/all) | haunted | 16:56:17 |
K900 | Nah that codebase is just absolutely insufferable | 16:56:54 |
dotlambda | Can you override filelock just for paperless-ngx? | 16:57:11 |
K900 | Well not without doing it for the entire package set it's built from | 16:57:22 |
dotlambda | So something propagates it? | 16:57:41 |
K900 | Yeah | 16:57:45 |
K900 | Sorry it's not setuptools that explodes, it's virtualenv | 16:58:06 |
K900 | But there's a weird middle version they can agree in | 16:58:13 |
dotlambda | I'm excited about paperless slop 3.0 | 16:58:14 |
K900 | Redacted or Malformed Event | 16:58:15 |
K900 | They tagged a beta of that | 16:58:46 |
K900 | I wonder if that one works | 16:58:50 |
Lach | I'm amazed how python ecosystem is still alive with no lockfiles having breaking changes in packages such as numpy | 17:04:37 |
dotlambda | Lockfiles are poison | 17:04:56 |
K900 | Oh no, there's plenty of lockfiles | 17:05:33 |
K900 | They're just impossible to actually use downstream | 17:05:43 |
Lach | Lockfiles enable proper dependency resolution and not the requirements.txt bs | 17:06:02 |
K900 | The Python ecosystem problem is not lockfiles | 17:06:18 |
Lach | They are impossible to use in upstream most of the time either D: | 17:06:38 |