Wiki article outlining common procedures for extending Scryer Prolog

[mthom]

To welcome and encourage contributions on both the system and library sides, it might help if the repo hosted a few wiki articles explaining how to extend Scryer Prolog. For example, the process of adding a new SystemClauseType variant ought to be documented, as it's almost always how Rust code is executed from the Prolog interpreter.

[triska]

How can we add a page to the Wiki?

Currently, when I click on https://github.com/mthom/scryer-prolog/wiki, I get redirected to https://github.com/mthom/scryer-prolog, i.e., to the main repository.

[mthom]

I've added a page. I suppose you can clone it locally, make edits, and then submit a PR as one does for commits?

[triska]

Thank you! We can clone it and edit it. However, pull requests are not supported for the Wiki.

I found the following workflow description:

https://stackoverflow.com/questions/10642928/how-to-pull-request-a-wiki-page-on-github

Contributors should push the changes to their own repository, and then file an issue or discussion with Scryer Prolog so that you can pick up the changes.

[panasenco]

If possible, could we put these instructions as comments in the Rust source code of SystemClauseType and then use docs.rs to generate a complete documentation site for the project?

We use a wiki site (Confluence) to describe technical objects at my work, and we ended up with half of all features completely undocumented, and the other half an incomprehensible mess. I'm actually spending a lot of my time at work rolling out a new automated documentation site to replace our wiki completely. The big advantage of automated documentation generation is that it documents everything and lays it out in a consistent fashion that's easy to understand and navigate.

[dcnorris]

This could help make an interesting development in the precautionary package, which already implements Rust numerics for continual reassessment method (CRM) dose-finding designs. Most of the 'action' in dose-finding is in ℤ, but many 'model-based' designs pass through ℝ on the way from discrete inputs (categorical toxicity assessments at discrete doses) to discrete decisions. A Scryer library that encapsulated these 'passages' through ℝ could enable all of (what precautionary is revealing as) the truly interesting aspects of the common dose-finding methods to be analyzed in Prolog.