Background and rationale

Last modified: 2026-05-10 13:19av bruker: Erik Hagen ID: 4fe26ad6-aa99-46b9-b522-b5f8bfa09bf7 SAMT-X/solution-samt-bu-docs

This section explains the needs that motivated the development of SAMT-BU Docs, the alternatives that were evaluated, and why we chose to build a custom solution.

The need

SAMT-BU is a cross-agency, cross-tier collaboration involving municipalities, directorates and county authorities. The documentation platform is intended to actively support this collaboration:

  • Non-technical subject-matter experts – lawyers, information architects, service designers and municipal staff – must be able to read, comment and contribute content without knowing Git or command-line tools
  • External contribution – people without direct repository access must be able to suggest changes that are then reviewed and approved by the project group
  • Bilingual interface – both the website and the editor interface must work in Norwegian and English, to support international collaboration

Alternatives evaluated

We evaluated two categories of solutions: wiki platforms and git-based CMS solutions.

Wiki platforms

Wiki platforms are a natural candidate for documentation with many contributors. We evaluated two variants:

MediaWiki is the software behind Wikipedia. It is mature, well-documented and open source. Semantic MediaWiki (SMW) is an extension that adds structured, machine-readable properties to wiki pages – think RDF-like relations and cross-page queries. Given that SAMT-BU works with interconnected services with rich metadata needs, and has a dedicated team-semantics, this is particularly relevant to consider.

Functional comparison:

FeatureHugo + Git (chosen)MediaWikiSemantic MediaWiki
Hierarchical navigation
Multilingual content✅ (Translate extension)
Web-based editing for non-technical users✅ (TipTap)✅ (VisualEditor)
Contribution flow without direct access✅ (PR-based)⚠️ (discussion + approval)⚠️
Version control and history⚠️ (git, not exposed to users)✅ (built-in, user-friendly)
Stable links that survive renaming⚠️ (to be built)✅ (page ID redirects, automatic)
Permalinks to specific versions⚠️ (to be built)✅ (oldid links, built-in)
Diff view between versions
Discussion and comments per page❌ (planned)✅ (talk pages)
Search✅ (Lunr.js)✅ + semantic queries
Semantic metadata and structured information⚠️ (frontmatter, manual)⚠️ (templates and categories)✅ (RDF properties, #ask)
Content from multiple repositories and teams✅ (Hugo Modules)⚠️ (interwiki, not seamless)⚠️
Custom visual profile⚠️ (possible, but complex)⚠️
Automated publishing and CI/CD⚠️ (not native)⚠️
Access control per section⚠️ (GitHub level)✅ (granular, built-in)
Bilingual editor interface⚠️ (possible, but requires customisation)⚠️

It is worth noting that several features we plan to build (stable links, version links, per-page discussion) are already available out of the box in MediaWiki and SMW.

Non-functional properties: MediaWiki/SMW requires a PHP server and database (MySQL/MariaDB), whereas Hugo generates static files. This is not in itself decisive, but it affects the operational model, hosting choices and integration with the project’s other tooling.

Git-based CMS solutions

Within git-based solutions, we evaluated three established tools:

ToolExternal contribution flowNorwegian/English UIAssessment
Decap CMS✅ Via users’ own GitHub forks❌ English onlyTested in practice. Non-technical users are confused by fork notifications and unexpected repositories appearing in their GitHub account. Too high a barrier.
TinaCMS / Tina Cloud✅ Via the vendor’s service account❌ English onlySolves the contribution problem, but is a paid SaaS service ($49+/month for editorial workflow) with a closed editor interface available only in English and no option to embed it in the site.
Keystatic❌ Not supported❌ English onlyDoes not support external contribution at all – requires write access for all users.

None of the alternatives covered all requirements simultaneously.

Chosen approach and trade-offs

We built a custom solution based on git and Hugo:

  • Built-in editor directly within the site’s pages – no external portal, no redirection to a separate interface
  • Worker-based contribution flow – a Cloudflare Worker creates a branch, commit and pull request on behalf of the contributor via a dedicated bot account. Contributors do not need write access to the repository
  • Bilingual editor interface – all text in the editor dialogs is available in Norwegian and English, controlled by the site’s active language
  • Content from multiple repositories – Hugo Modules pulls content from team-architecture, team-semantics and other repositories into one unified site
  • Open source, no vendor lock-in – we own the solution and can develop it freely

The primary argument against MediaWiki/SMW was not technical but organisational: the project’s existing workflow is already git- and GitHub-based, and a wiki platform would have introduced a parallel system with a different user model and ownership structure. The federation of content from multiple repositories via Hugo Modules – which reflects the project’s team structure – also has no good equivalent in MediaWiki.

The areas where MediaWiki/SMW is stronger (version history exposed to users, stable links, per-page discussion) have been identified as gaps and are planned to be addressed over time.

A distinctive advantage: bilingual support

None of the tools we evaluated offer a fully multilingual editor interface. For SAMT-BU Docs, this is a significant advantage: Norwegian users work in Norwegian, while international partners – for example, those participating in initiatives such as Skills Dataspace in the EU – can use the same interface in English.

Further reading