Dokmelody

I restarted working on Dokmelody in the last two weeks.

I studied a lot of papers and personal notes, trying to synthesize a design in org-roam scratchpad. The approach I’m following is:

• read a paper about a problem to solve and the proposed solution
• check if current Dok design can solve the problem, otherwise try to integrate the idea of the paper in the Dok design

There are still a lot of notes to integrate, so I will continue also in next weeks.

I spent too much time rethinking the organization of the websites. This is an unacceptable yak shaving for an alpha-state project.

I switched from Git and Codeberg to Pijul and Nest. Some reasons are:

• Pijul supports merges and forks of a codebase in a better way respect Git, because it remembers the merge operations done. This feature fits well with the ADR-004: Favors branching and forking of Dokmelody.
• I want to learn Pijul, and maybe I like to use Pijul “theory of patches” into some parts of Dokmelody system, for managing the forks of data between different communities.
• There are future plans, for a self-hostable Nest. Something like Fossil.
• Nest has also a “discussion” section.

I’m using DITA for Dokmelody documentation. The reasons are:

• navigation in documentation is very easy
• favours the use and reuse of chunks of information

I switched to MIT License. The reasons are described in ADR-003: Use MIT and CC-BY-4.0 licenses:

I decided to use a redo build system, respect a traditional “make-like” approach. A make-file is essentially a bottom-up specification of how to build the system starting from initial source files. A “redo” approach is a top-down specification of how to build a product, checking if some parts do not need an update. This approach is more similar to normal code, because every part of a build is like a function that can call other functions.

I’m developing in a private branch a new version of Dok language syntax and semantic. Next week, I hope to release something of acceptable.

I want to release a proof of concept before the end of 2023. I will document the progress every week as a source of motivation.

This week, I published https://www.dokmelody.org. As a form of yak shaving, I created also too many websites and services, that there will be never used for the foreseeable future.

Whenever possible, I selected federated and distributed tools. I discarded the previous Fossil installation, despite it is a very nice tool, because it is not federated, while Gitea/Codeberg are adding activity-pub support.