Design history¶
Every change to the solution was planned in a new version with its requirements, acceptance checks and the evidence behind each decision. Newest first.
| Plan | Status |
|---|---|
| Plan v7: 2 vCPU / 8 GB target, latest LTS only, latest stable dependencies | Implemented; validated by the Ubuntu 26.04 go-ahead gate on the simulated 2 vCPU / 8 GB machine (see tests/results/*_ubuntu-26.04/). Supersedes plan-v6.md. |
| Plan v6: semantic versioning and releases | Superseded by plan-v7.md (2 vCPU target, latest LTS, docling v1.36.0). Supersedes plan-v5.md. |
| Plan v5: benchmark-driven defaults, test suite, open-source release | Superseded by plan-v6.md (semantic versioning). Supersedes plan-v4.md. |
| Plan v4: PDF → Markdown extraction via docling-serve (as implemented) | Superseded by plan-v5.md (scanned PDFs no longer split; timeout per page). Supersedes plan-v3.md. |
| Plan v3: PDF → Markdown extraction via docling-serve | Superseded by plan-v4.md. Supersedes plan-v2.md. |
| Plan v2: PDF → Markdown extraction via docling-serve | Superseded by plan-v3.md. Supersedes plan-v1.md. |
| Plan v1: PDF → Markdown extraction via docling-serve | Superseded by plan-v2.md |
Proposals (not implemented)¶
- Proposal: Amazon S3 (via S3 Files) as an optional input provider: Proposal for discussion, not implemented. It is part of the design thinking for the next stage of the solution. Local disk (EBS) stays the default and the only validated option.
Landing page plans¶
| Plan | Status |
|---|---|
| Landing page: Plan v2 (GitHub Pages, with local Pages simulation) | Built and checked locally (LA1, LA2, LA4, LA5, LA7–LA11 pass); publishing waits for the repository. Supersedes landing-page-plan-v1.md. This is a separate track from the extractor plans. |
| Landing page: Plan v1 (GitHub Pages) | Superseded by landing-page-plan-v2.md (local Pages simulation, link handling). This is a separate track from the extractor plans (plan-v1 … plan-v5). |