Scheduled posts from timeline yaml

[andrewpbray]

I have a blog that I'd like to publish on a schedule, with the publishing dates stored in a file that's separate from the posts. It seems like the scheduling question would be well-addressed by the filter described in #3114. What is the best approach for propagating the date information into the documents or into the render process?

I've put together a repo that explores a couple of different approaches to this.

https://github.com/andrewpbray/scheduled-posts-options

They're described in the README and I've copied them here below. It strikes me that this is a version of a more general question: what's the best way to render documents subject to document-specific metadata that is held in a separate file?

The three approaches below reference a _schedule.yml that would look something like this:

schedule:
  - filename: "topic-1"
    date: "01/01/22"
  - filename: "topic-2"
    date: "01/02/22"
  - filename: "topic-3""
    date: "01/03/22"

1. YML injection via .md file created in pre-render script

Write a pre-render script that parses _schedule.yml and creates a tiny _publish-date.md file next to each post that adds to the metadata via an include. This is quasi-implemented in Topic 2.

Downsides:

1. feels kludgy
2. generates a bunch of files, though this could be addressed by
   • adding files to .gitignore
   • keep files in dedicated include directory
   • create a post-render script to clean out that directory

2. Reference _variables.yml in document YAML

Create a _variables.yml file that includes the publish dates for each post. That variable is then referenced in the yaml header of the post. This is implemented in Topic 3 (though apparently I've formatted the date field incorrectly).

A similar use case is discussed here.

Downsides:

1. Requires tailoring the yaml of each doc to reference the appropriate date.
2. Seems like a misuse of variables. These are meant to be shared across documents, right? Here they're document specific.

3. Write a filter

Changing the meta data of a documents seems solidly within the functionality of a filter. The wrinkles that I see here:

1. The filter will need to reference information in _schedule.yml. Is there an example of a filter that does this that I could look at?
2. Right now, _schedule.yml links the publish date with the file / directory name of the post. I'm not sure how to reference that information in the filter. Would it make more sense to reference the title field in the yaml?

[srvanderplas]

This would also be very useful for e.g. course websites, where you don't want to overwhelm students at the beginning of the semester (so you only want content that is initially relevant) but you have the whole course prepped ahead of time. It would also make changing dates for an entire semester at once much easier.

[andrewpbray]

@srvanderplas I figured out a way to implement this using R scripts that are run before I run quarto render. The repo of course materials is here and the file that controls the publish dates is called _course-settings.yml. I render the course website through github actions found here, which runs this script before rendering.

The key idea of the script is to read the yml file and prepend _ to any files with publish dates in the future. This has the effect of ignoring these files during the render process.

[mcanouil]

You could also add the files in. project.render using negation, see https://quarto.org/docs/websites/index.html#render-targets.
This means only one file to modify.

[andrewpbray]

I believe when you use that approach, the ignoring happens after quarto builds the document listings. That means that pages like the problem sets and labs have blank spots.