Open Questions
Kept deliberately short. If this list grows past ~15 items, we’re accumulating questions faster than we’re resolving them — slow research down and resolve some before adding more.
Current questions
Section titled “Current questions”Q01 — What’s in the local Obsidian research vault?
Section titled “Q01 — What’s in the local Obsidian research vault?”Asked: 2026-04-11 Source: Existing Work audit Why it matters: we may be re-researching things that already exist in prior notes in the local Obsidian research vault (local research vault). Blocks: efficient use of Phase R time. Status: needs a manual audit pass.
Q02 — Is astro-loader-obsidian sufficient for Tier 1 translation?
Section titled “Q02 — Is astro-loader-obsidian sufficient for Tier 1 translation?”Asked: 2026-04-11 Source: Translation Layer Why it matters: determines whether the translation layer is a “wire up existing plugins” problem or a “write custom plugins” problem. Changes the scope of Phase 1 by 5-10x. Blocks: Phase 1 translation layer spike. Status: unknown. Needs hands-on testing with a vault of real Obsidian-flavored content.
Q03 — Should the Web CMS show rendered Obsidian preview, or raw markdown?
Section titled “Q03 — Should the Web CMS show rendered Obsidian preview, or raw markdown?”Asked: 2026-04-11 Source: Contribution Workflows Why it matters: determines whether Decap needs a custom preview pipeline that can render Tier 1 features, or whether we accept “raw markdown with syntax highlighting” as the editing surface. Blocks: Web CMS design in Phase 2. Status: depends on how Decap / Sveltia / Tina each handle custom preview components.
Q04 — How do we guarantee no lossy round-trip in practice?
Section titled “Q04 — How do we guarantee no lossy round-trip in practice?”Asked: 2026-04-11 Source: Translation Layer, Principle 2 Why it matters: Principle 2 is only a principle if it’s enforceable. Otherwise it’s an aspiration. Blocks: the claim of “round-trip editability” in vision and principles. Status: probably property-based tests on a vault fixture corpus, but the framework for this doesn’t exist yet.
Q05 — What’s the incremental build story for a vault of thousands of pages?
Section titled “Q05 — What’s the incremental build story for a vault of thousands of pages?”Asked: 2026-04-11 Source: Architecture, carried forward from archived Phase-0 docs Why it matters: if every edit triggers a full rebuild of a 1000-page vault, the feedback loop breaks and contributors lose trust in the pipeline. Blocks: scaling, but not Phase R itself. Status: Astro 5+ has improved incremental support; needs benchmarking on a realistic vault.
Q06 — Content addressability: slugs, hashes, or title-based with redirects?
Section titled “Q06 — Content addressability: slugs, hashes, or title-based with redirects?”Asked: 2026-04-11 Source: Primitives · Content-Addressable URLs Why it matters: URL stability is a cardinal-sin property for a wiki. External links must survive vault reorganization. Blocks: any production deploy. Status: no decision yet. Three candidates: explicit frontmatter slug, content-hash, or auto-slug with a redirect table.
Q07 — How does the OAuth proxy for Decap get hosted?
Section titled “Q07 — How does the OAuth proxy for Decap get hosted?”Asked: 2026-04-11 Source: Architecture Why it matters: Decap’s Open Authoring needs an OAuth proxy so the GitHub OAuth client secret isn’t exposed in the browser. The proxy is a few lines of code but needs a hosting target. Blocks: Web CMS setup in Phase 2. Status: probably Cloudflare Pages Functions (matches sibling pattern); needs verification that Functions supports the redirect dance.
How to resolve a question
Section titled “How to resolve a question”- Spend the research time (could be a session, could be a sprint)
- Write the answer and rationale in Trade-offs or in the relevant design page
- Move the question out of this file into a “Resolved Questions” log with date + answer + what-it-unblocked
- If the answer reveals deeper questions, those become new entries here
The dirt house rule: a good answer creates two new questions, and that’s fine. Better to have precise questions than vague confidence.