12 Jul 2021 |
matthewcroughan - nix.zone | Wow that's fast. | 14:31:47 |
matthewcroughan - nix.zone | huh, I wonder if this is in the LSP? | 14:32:06 |
matthewcroughan - nix.zone | I don't think it is, but this could seriously enhance the existing lsp. | 14:32:32 |
niksnut | That's what we used to do right? (Read the entire options.json / packages.json into the client.) Problem is that it doesn't scale very well. | 15:15:21 |
andi- | In reply to @niksnut:matrix.org That's what we used to do right? (Read the entire options.json / packages.json into the client.) Problem is that it doesn't scale very well. That isn't what my approach does. I only have ship a limited search index at first and as you type lazy-load more detailed data. | 15:16:38 |
andi- | In reply to @niksnut:matrix.org That's what we used to do right? (Read the entire options.json / packages.json into the client.) Problem is that it doesn't scale very well. * That isn't what my approach does. I only ship a limited search index at first and as you type lazy-load more detailed data. | 15:16:48 |
andi- | That page is just a POC of the idea. It could be more efficient or match better but I didn't spent more time on it. | 15:17:52 |
matthewcroughan - nix.zone | Yeah I noticed that my enter key wasn't the thing triggering the search. | 15:23:47 |
matthewcroughan - nix.zone | And it is lazy loaded, as it should be, it's brilliant. | 15:23:56 |
matthewcroughan - nix.zone | Have you got the sources for how this was done? I have never implemented anything like that before and wonder how hard it could be. | 15:24:20 |
andi- | In reply to @matthewcroughan:defenestrate.it Have you got the sources for how this was done? I have never implemented anything like that before and wonder how hard it could be. Yeah, it is mostly just using lunr and hacking around it to make it useful for that POC https://github.com/andir/nixos.dev-option-search/blob/master/create-index.js | 15:28:35 |
20 Jul 2021 |
| sumner left the room. | 21:43:02 |
21 Jul 2021 |
samueldr | another person also had a similar PoC with the package search | 18:39:51 |
samueldr | (shared when we were already committed to going with ES) | 18:40:03 |
samueldr | not sure how/where it differs | 18:40:11 |
samueldr | I've been using the options search a lot these past few days and... not been pleased with the irrelevance of the "relevant" results :( | 18:40:42 |
samueldr | garbas: reminder that in 19 minutes there's that bi-weekly call | 18:41:30 |
samueldr | (or not?) | 19:01:03 |
samueldr | (there is) | 19:18:27 |
samueldr | (there was) | 20:18:14 |
tomberek | samueldr: this is what I meant by a non-invasive way to try out Nix: https://github.com/nix-dot-dev/nix.dev/pull/169 | 20:33:42 |
samueldr | nice! | 20:34:21 |
matthewcroughan - nix.zone | Huh I didn't realise you could contribute to nix.dev. That is certainly cool. Though I believe sphinx/rst are the wrong format to use, I may set up an alternative that uses MkDocs. MkDocs lets you use templating (jinja) and python to change the page contents, from within the MD file so you can have smart pages that have logic in them. Like if page.metadata == arm64 then include ../common/arm64.md | 20:42:37 |
samueldr | ugh, please, not yet another source of documentation, it's already too fragmented | 20:43:35 |
matthewcroughan - nix.zone | I was looking for evidence of automation/coding in the official Nix docs, and it looks like it's 100% hand written? | 20:43:43 |
samueldr | "official docs" is still un flux because of RFC#whatever the one for markdown | 20:44:08 |
samueldr | * "official docs" is still in flux because of RFC#whatever the one for markdown | 20:44:15 |
matthewcroughan - nix.zone | I just don't think a lot of the stuff that is handwritten can be maintained. | 20:44:19 |
matthewcroughan - nix.zone | Well it can, with dedication! But it's harder than it has to be. | 20:46:05 |
andi- | Isn't everything apart from API docs manual writing anyway? How can you automate that and nix.dev is mostly not api docs IIRC. | 20:46:59 |