| 24 Apr 2024 |
|  @gibbs_ensamble:matrix.org left the room. | 21:16:41 |
| 25 Apr 2024 |
|  SomeoneSerge (hash-versioned python modules when) changed their display name from SomeoneSerge (void) to SomeoneSerge (UTC+1). | 23:01:13 |
| 26 Apr 2024 |
|  Bruno Rodrigues changed their profile picture. | 09:19:24 |
| 27 Apr 2024 |
|  @ygt:matrix.org joined the room. | 18:21:19 |
| 29 Apr 2024 |
|  @anjoletto:matrix.org joined the room. | 01:46:16 |
| SomeoneSerge (hash-versioned python modules when) changed their display name from SomeoneSerge (UTC+1) to SomeoneSerge (is taking time off and doesn't want to hear about it). | 11:43:13 |
|  NixOS Moderation Botchanged room power levels. | 15:29:11 |
| 30 Apr 2024 |
|  ondt joined the room. | 22:18:47 |
| 1 May 2024 |
| NixOS Moderation Botchanged room power levels. | 15:06:23 |
|  @daschw:matrix.org joined the room. | 18:52:42 |
| 3 May 2024 |
|  @kyaru:ptt.moe joined the room. | 19:24:03 |
| 4 May 2024 |
| SomeoneSerge (hash-versioned python modules when) changed their display name from SomeoneSerge (is taking time off and doesn't want to hear about it) to SomeoneSerge (Way down Hadestown). | 21:03:47 |
| 5 May 2024 |
|  @markus.lohmayer:matrix.org joined the room. | 06:55:04 |
| @markus.lohmayer:matrix.org | Sorry in advance if my question doesn't fit in here. I would like to build a CI workflow for reproducible academic work. Basically the main document, written in LaTeX includes certain plots, which would be costly to recreate on every run. Hence, the workflow should recreate those only if the code that produced them changed. Similarly, the LaTeX should only be recompiled if the actual figures changed (as not all code changes would lead to different figures). Is it difficult to build such CI workflows? | 06:59:27 |
| @markus.lohmayer:matrix.org | * Sorry in advance if my question doesn't fit in here. I would like to build a CI workflow for reproducible academic work. Basically the main document, written in LaTeX includes certain plots, which would be costly to recreate on every run. Hence, the workflow should recreate those only if the code that produced them changed. Similarly, the LaTeX should only be recompiled if the actual figures changed (as not all code changes would lead to different figures). Is it difficult to build such CI workflows? (nor sure if it matters too much, but code is written in Julia) | 06:59:57 |
| Bruno Rodrigues | I do something like that using R but I rebuild the whole thing as there's no computation involved | 15:02:04 |
| Bruno Rodrigues | here's a demo https://github.com/b-rodrigues/render_quarto_test | 15:02:12 |
| Bruno Rodrigues | but if I only wanted to rebuild parts of it, I'd use targets instead of calling quarto directly | 15:03:13 |
| Bruno Rodrigues | Also, this uses a quarto markdown source document, but you can just as well use Latex | 15:03:52 |
 guangtao | https://github.com/hardenedlinux/AISecurity-Research-Template | 15:24:07 |
| guangtao | quarto+jupyenv+drvparts+dream2nix+std example heee | 15:24:57 |
| @markus.lohmayer:matrix.org | Thanks for sharing your templates. While I can definitely learn something from each of them, I am still wondering how I could compile the plots if the code changed and compile the LaTeX if the TeX code or the plots changed. | 17:34:27 |
| @markus.lohmayer:matrix.org | I am thinking if this is something that should be done with Nix. Maybe not since Nix puts everything it builds into the Store, right? | 17:35:44 |
| @markus.lohmayer:matrix.org | Isn't this a problem that comes up eg with monorepos? isn't there are simple solution then? | 17:36:44 |
| guangtao | In reply to @markus.lohmayer:matrix.org
Thanks for sharing your templates. While I can definitely learn something from each of them, I am still wondering how I could compile the plots if the code changed and compile the LaTeX if the TeX code or the plots changed. I don't really understand your motivation. if the plots changed just quarto rebuild the qmd dir, that's all | 17:56:04 |
| guangtao | nix only used to build and control the kernels' generatation | 17:56:38 |
| guangtao | also there is no reason to exec 'quarto build' inside the nix store | 17:57:46 |
| @markus.lohmayer:matrix.org | OK, I don't know quarto, that's the point. but it sounds like it is doing what I need. I will have a look | 18:21:48 |
| @markus.lohmayer:matrix.org | Ok, I already stated that I am stuck with LaTeX, so suggesting quarto, as I understand by looking at quatro.org, is not helpful. | 18:26:47 |
 jbedo | fetchurl and fetchgit have postFetch which you could use to extract the paper/figure code | 22:10:11 |
