| 24 Oct 2024 |
 w | * Folks, is there a group for nignx ou webservers? | 16:00:52 |
| 25 Oct 2024 |
 Pol | In reply to @ma27:nicht-so.sexy
no promises, may be able to take a look tonight. Feel free to ping me again if you haven't found a solution until then. Gentle ping ^^ :) | 08:16:28 |
 ma27 | uh yes.
didn't forget, but fell asleep 🙈
let's retry this afternoon! | 08:30:02 |
| Pol | Haha that's a good illness, happen all the time to me too :D | 08:40:46 |
| ma27 | Pol hansemschnokeloch would you mind sharing the state that causes the configure error? It's kinda hard to follow what you've changed so far and I don't want to debug something unrelated 😅 | 09:41:56 |
| Pol | We wish we could fix the issue here: https://github.com/NixOS/nixpkgs/pull/350771 | 09:43:47 |
| Pol | I wish we could pass the php in the top-level parameters and modify the php derivation. | 09:44:22 |
| Pol | But it's not possible sadly. | 09:44:28 |
| Pol | I'm trying to find an elegant solution | 09:44:38 |
| Pol | But I'm running out of ideas. | 09:44:44 |
| ma27 | ah I misread it as debugging the error from above. OK sure, will take a look at that then! | 09:44:57 |
| Pol | Merci <3 | 09:45:18 |
 Jan Tojnar | Pol: I would not override PHP, I would just mark it as broken without ZTS | 09:48:25 |
| Jan Tojnar | otherwise it will probably result in symbol conflicts or other confusing issues at runtime when using non-ZTS package | 09:50:02 |
| Pol | This is another way to see this indeed. So we let the user customize PHP, it won't be done by the extension if I understand correctly? | 09:51:03 |
 hansemschnokeloch | If I understand it correctly, I just add meta.broken = !php.ztsSupport; ? | 10:00:25 |
| Pol | Yes I think that's what he means. | 10:01:44 |
| ma27 | Hmm, after thinking a bit abotu this, I agree with Jan. | 10:02:27 |
| ma27 | the current state of the PR builds because we provide the correct PHP.
But if you do php.withExtensions [ps.parallel], it would build fine, but just wouldn't work, correct? | 10:03:21 |
| ma27 | so yes, I think a user should explicitly "declare" a PHP with ZTS and only then zts-requiring extensions can be added. | 10:03:47 |
| ma27 | does that make sense? | 10:03:56 |
| hansemschnokeloch | It make sense to me, I first thought of adding an assertion somewhere like here, but didn't figured out how to do it in a php extension https://github.com/NixOS/nixpkgs/blob/86e1ad4ec007f4f0e9561886935fe9b278860de8/nixos/modules/services/web-servers/apache-httpd/default.nix#L637 | 10:07:12 |
| Pol | That could also be an option | 10:09:04 |
| Pol | (maybe I'm wrong too) | 10:09:16 |
| ma27 | I'd rather do it on a package-level than module-level since this is a problem on the package-level. | 10:09:41 |
| ma27 | the code you've linked is correct because it's an issu ein the combination of httpd and php (AFAIU) | 10:09:54 |
| ma27 | I mean, otherwise we catch the problem for whatever fpm/httpd/? service we build this into, but not for deployments that only use the packages, but run them on their own (i.e. without services). | 10:10:52 |
| Pol | In reply to @drupol:matrix.org
That could also be an option No, I mean, we could use an assertion like the one linked up there in the package parallel too. | 10:11:40 |
 Shyim | can we put php extension in pkgs/by-name or is that not working? | 16:34:37 |
| Pol | I don't think this is such a such a good idea | 19:41:53 |
