How best to serve versioned files?

[matt-dray]

What is the best way to version files in this repo and refer to them from other repos?

For example, let's imagine a Shiny app that contains a reference file. The app and the file are in the same repo and therefore they are the same version. If the reference file is moved into this repo, the app and the file versions will become decoupled. Independent changes may happen in the app or to the file, so a new release of the app should be tied to the concurrent version of the reference file.

We serve the json schema from nhp_models with a branch per version. Could something similar be achieved here? There could be a lot of files to serve, so it may not be viable.

[yiwen-h]

I guess my questions are:

• How many files?
• How often will they be updated?
• If we update the file, is it important that the old version is still viewable/accessible

I think any solution should be as simple as possible.

Could we just use GitHub for versioning? older versions of the file will still be available in the tagged commit if we need to refer back but we should always try and use whatever is on main wherever possible? Then the file version would be linked to the version of the app (we could track this in release notes?)

[tomjemmett]

worth considering using BI techniques like SCD Type 2? This works well if data is tabular and doesn't change in structure over time