Skip to content

Reorder detail_design.md so the architecture diagram leads Project Structure#13

Merged
napinoco merged 4 commits into
mainfrom
claude/claude-md-docs-structure-r5rpg8
Jul 7, 2026
Merged

Reorder detail_design.md so the architecture diagram leads Project Structure#13
napinoco merged 4 commits into
mainfrom
claude/claude-md-docs-structure-r5rpg8

Conversation

@napinoco

@napinoco napinoco commented Jul 7, 2026

Copy link
Copy Markdown
Owner

Move Component Architecture (with the process-flow diagram) ahead of
Project Structure in detail_design.md so readers see the system's
behavior before the directory tree. Rename basic_design.md's "Core
Architecture" heading to "Deployment & Execution Overview" to avoid
confusion with detail_design.md's differently-scoped diagram, and add
a link from README.md's Architecture section to the detailed diagram.

claude added 4 commits July 5, 2026 14:55
…ructure

Move Component Architecture (with the process-flow diagram) ahead of
Project Structure in detail_design.md so readers see the system's
behavior before the directory tree. Rename basic_design.md's "Core
Architecture" heading to "Deployment & Execution Overview" to avoid
confusion with detail_design.md's differently-scoped diagram, and add
a link from README.md's Architecture section to the detailed diagram.
…ides

README's Documentation section now points to docs/guides/, which was
previously unreachable from the two primary entry points (README.md,
CLAUDE.md).

LOCAL_DEVELOPMENT_GUIDE.md: replace the Quick Start/Installation/Running
Benchmarks sections (near-duplicates of README's Quick Start, and stale:
wrong Python version, missing --recursive clone, wrong docs/ output path,
fabricated solver/problem counts) with links to README and detail_design.md.
Fix report file names and paths in "Understanding the Output" to match
actual output (docs/pages/*.html), and link data-file details to
EXPORT_GUIDE.md instead of duplicating them.

CONFIGURATION.md: drop a fabricated "expected output" block with stale
counts, and repoint the solver-addition cross-reference to
detail_design.md now that LOCAL_DEVELOPMENT_GUIDE.md no longer duplicates
that section.

EXTERNAL_LIBRARIES.md: remove emoji headers, fix submodule clone commands
and problem-registry example to match the actual .gitmodules remotes and
config/problem_registry.yaml schema (the previous `external_libraries:`
YAML example didn't match the real `problem_libraries:` flat schema), and
replace a fabricated "Adding New External Libraries" walkthrough with a
link to detail_design.md's actual loader registration steps.
NETLIB (94 LP problems, MPS format) was fully integrated in a prior
change (MPS loaders for both Python and MATLAB, benchmark results
already recorded) but was never added to README.md, CLAUDE.md,
basic_design.md, or detail_design.md, which still only described
DIMACS and SDPLIB. Add it alongside the existing libraries in the
problem library lists, directory trees, the Mermaid diagram, the
problem registry schema comment, and the registered-problem count
(roughly 215 total, not ~120).

Also unify the minimum MATLAB version across docs: setup_matlab_solvers.m
is the authoritative source (R2020a or newer, matching README.md);
CLAUDE.md said R2024+ and docs/guides/CONFIGURATION.md said R2019b.
Both now read R2020a+.
…TLIB loader bug

NETLIB_KNOWN_OBJECTIVES.md existed as the only per-library data-provenance
note in docs/guides/, which read as an unexplained asymmetry. DIMACS
(34/37 problems) and SDPLIB (86/86) also carry known_objective_value
entries, so add the equivalent source/caveats note for them
(DIMACS_SDPLIB_KNOWN_OBJECTIVES.md), sourced from the DIMACS and SDPLIB
submodules' own READMEs. This also documents the historical sign-convention
fixes (commits b8b5ec2, cea8c62) and two caveats not previously recorded
anywhere: hinf12/hinf13's known values are flagged "(?)" (uncertain
accuracy) by the DIMACS source, and fap25's value is a lower bound, not
a proven optimum.

Separately, NETLIB_KNOWN_OBJECTIVES.md documents a real, currently unfixed
bug (mps_loader.py doesn't reorder free variables, giving wrong objective
values for 11 NETLIB problems under Python solvers) that was only
discoverable by reading a guide nothing links to. Surface it in README.md's
development-status banner and add a "Known Issues" section to CLAUDE.md so
it isn't missed before trusting or building on NETLIB Python-solver results.
@github-actions

github-actions Bot commented Jul 7, 2026

Copy link
Copy Markdown

📊 Preview Reports Ready!

Your preview has been deployed and is available at:

🔗 View Preview Reports

Preview Details

Available Reports

Notes

  • Preview uses the pre-built reports from your branch
  • Preview will be updated automatically when you push new commits
  • Preview will be cleaned up automatically when the PR is closed

Preview deployment powered by GitHub Actions

@napinoco napinoco merged commit a09b159 into main Jul 7, 2026
2 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants