| 14 Jan 2024 |
 danielsidhion | In reply to @johannes.kirschbauer:scs.ems.host
Is there any tutorial how to develop on nixos-render-docs?
https://github.com/NixOS/nixpkgs/tree/528a4bb0ac5e5af73445472ec25945460add7768/doc#contributing-to-this-documentation
The only sentence there is:
The rendering tool is nixos-render-docs, sometimes abbreviated nrd.
I am not super familiar with the implied workflow python development on that package.
I'm Calling nix develop .#nixos-render-docs to load the build dependencies of the package in my shell.
But then executing pytest
Results in an error
ImportError while importing test module 'nixos-render-docs/src/tests/test_asciidoc.py'.
....
import nixos_render_docs as nrd
E ModuleNotFoundError: No module named 'nixos_render_docs'
Any ideas ?
Just spent some time looking into this. You need to run python -m pytest instead of pytest because of this: https://docs.pytest.org/en/7.4.x/explanation/pythonpath.html#invoking-pytest-versus-python-m-pytest | 20:54:42 |
| 15 Jan 2024 |
 @mightyiam:matrix.org | Update: the Nix doctester is still being worked on at four hours per week. | 15:11:33 |
 Chris McDonough | I've made the changes to https://github.com/NixOS/nixpkgs/pull/277534 that we talked about during today's docs meeting, hopefully mergeable at this point when CI is done. | 16:32:27 |
| Chris McDonough | Redacted or Malformed Event | 19:15:50 |
| Chris McDonough | * The rebuild impact of https://github.com/NixOS/nixpkgs/pull/275701 (the patch by cwp that fixes Python binary wrappers) is pretty huge. About half of everything depended upon in the devenv examples must be recompiled. Everything works so far, but I can imagine pushing that PR through into nixpkgs master is going to take a good long while due to its impact. I think maybe we should think about using our own makeWrapper in devenv itself that produces fixed Python binary wrappers in the meantime? Basically just kinda copy over the stuff from python/wrapper.nix and use it instead of the stuff in nixpkgs? | 19:16:37 |
| danielsidhion | Is there a preferred manpage resource that I can use to link directly to the manpage of a command inside the nixpkgs manual? I did a search on the manual, but only found text like ...can be found in the dyld(1) manpage without any links to any manpage. | 22:01:45 |
| danielsidhion | Arch has e.g. man.archlinux.org which they use a lot in their docs to link to manpages. I'm wondering if there's something similar for nixpkgs/nixos that I can use. Otherwise I'll just keep it without a link | 22:02:41 |
 @bzzm3r:matrix.org | In reply to @danielsidhion:nixos.dev
Arch has e.g. man.archlinux.org which they use a lot in their docs to link to manpages. I'm wondering if there's something similar for nixpkgs/nixos that I can use. Otherwise I'll just keep it without a link I haven't run into any such thing yet, no. On the Discord people often end up linking to man.archlinux.org, or even wiki.archlinux.org | 22:08:58 |
| @bzzm3r:matrix.org | When googling for something in the man pages, there are a variety of hosts that come up | 22:09:30 |
| @bzzm3r:matrix.org | are there some out of those that are particularly reputable? | 22:10:14 |
 infinisil | danielsidhion: bzm3r: https://github.com/NixOS/nixpkgs/tree/master/doc#roles links to manpage-urls.json, seems to be mainly systemd links, I can't see much else | 23:00:35 |
| infinisil | So I guess the answer is that there's no default for now :) | 23:00:53 |
| infinisil | (though there might be some used in various other parts of the docs) | 23:01:08 |
| danielsidhion | Yeah, I didn't really know if there was some kind of default. The nixpkgs manual doesn't use that {manpage} syntax at all. Guess I'll start using it then. | 23:43:43 |
| danielsidhion | My concern isn't in finding some reputable source, but mostly in having a link to the manpages corresponding to the version of the package in nixpkgs | 23:44:34 |
| infinisil | Ah yeah that could be a problem. If we want to be fancy and rabbit-holey, we could make Hydra render all manpages from package builds and create some convenient redirects :D | 23:58:48 |
| 16 Jan 2024 |
| danielsidhion | I'd have to spend a lot of time to figure this out so I added it to my notes to look into this at some point. I think it would be a good improvement to have access to the manpages from nixpkgs directly | 00:02:44 |
| @bzzm3r:matrix.org | proofconstruction: do you usually run the Monday meetings? | 04:36:06 |
 @qyriad:matrix.org | In reply to @infinisil:matrix.org
Ah yeah that could be a problem. If we want to be fancy and rabbit-holey, we could make Hydra render all manpages from package builds and create some convenient redirects :D this would be awesome to have | 04:40:00 |
| @bzzm3r:matrix.org | fricklerhandwerk: from the language reference ("lang-ref", https://github.com/NixOS/nix/tree/master/doc), re: nix:
- *declarative*
There is no notion of executing sequential steps.
Dependencies between operations are established only through data.
let ... in blocks: are these staged/sequenced definitions?- similarly:
rec applied to an attribute set, is there implicit staging/sequencing of definitions?
- do data dependencies not imply staging/sequencing?
| 05:06:38 |
| @bzzm3r:matrix.org | I think I understand sufficiently now: "sequencing" here refers to sequencing of computing operations. There is no explicit sequencing of computing operations. Only implicit sequencing of computing operations through data dependencies. | 05:26:23 |
| @bzzm3r:matrix.org | * I think I understand sufficiently now: "executing sequential steps" here refers to explicit sequencing of computing operations. There is no explicit sequencing of computing operations. Only implicit sequencing of computing operations through specification of data dependencies. | 05:26:55 |
| @bzzm3r:matrix.org | * I think I understand sufficiently now: "executing sequential steps" here refers to explicit sequencing of computing operations. There is no explicit sequencing of computing operations. Only implicit sequencing of computing operations through specification of data dependencies is allowed. | 05:45:47 |
| @bzzm3r:matrix.org | * I think I understand sufficiently now: "executing sequential steps" here refers to explicit sequencing of computing operations. There is no explicit sequencing of computing operations. Sequencing of computing operations only happens implicitly through specification of data dependencies. | 05:46:09 |
 fricklerhandwerk | In reply to @bzzm3r:matrix.org
I think I understand sufficiently now: "executing sequential steps" here refers to explicit sequencing of computing operations. There is no explicit sequencing of computing operations. Sequencing of computing operations only happens implicitly through specification of data dependencies. That’s right, it’s not exactly right. Of course the evaluator does things in some sequence, you just don’t have much control over it. The “executing sequential steps” is an old fragment I took from something or someone. Feel free to suggest a rewording in a PR | 07:09:49 |
| fricklerhandwerk | In reply to @bzzm3r:matrix.org
I think I understand sufficiently now: "executing sequential steps" here refers to explicit sequencing of computing operations. There is no explicit sequencing of computing operations. Sequencing of computing operations only happens implicitly through specification of data dependencies. * That’s right, it’s not very precise and, if you think too hard about it, potentially misleading. Of course the evaluator does things in some sequence, you just don’t have much control over it. The “executing sequential steps” is an old fragment I took from something or someone. Feel free to suggest a rewording in a PR | 07:10:20 |
| fricklerhandwerk | Thinking too hard about stuff usually reveals that it’s all a scam… | 07:11:37 |
| @bzzm3r:matrix.org | Sounds good. Currently I'm working on a PR to enable auto-formatting of EBNF in the Nix (reference) manual. (Rather than manually formatting the markdown...) | 07:11:38 |
| fricklerhandwerk | In reply to @bzzm3r:matrix.org
Sounds good. Currently I'm working on a PR to enable auto-formatting of EBNF in the Nix (reference) manual. (Rather than manually formatting the markdown...) That would be cool. Just writing down the EBNF would be way cooler though because the first is cosmetic while the second is essential. | 07:12:48 |
| fricklerhandwerk | In reply to @bzzm3r:matrix.org
Sounds good. Currently I'm working on a PR to enable auto-formatting of EBNF in the Nix (reference) manual. (Rather than manually formatting the markdown...) * That would be cool. Just writing down the EBNF in the first place would be way cooler though because the first is cosmetic while the second is essential. | 07:12:58 |
