Update v1 templates for PyRel PR #730 API changes by chriscoey · Pull Request #28 · RelationalAI/templates

chriscoey · 2026-03-19T07:58:02Z

Summary

Migrates all 22 prescriptive templates to the new result accessor API from PyRel PR #730 (available in PyRel v1.0.6)
Adds engine-side model.require() ICs for solve status validation — avoids querying data to the client
Adds p.verify() constraint checking pattern to shift_assignment
Fixes demand_planning_temporal README {{name:float}} → {Float:name} Property syntax

API migration (all 22 templates)

Before	After
`p.display_solve_info()`	`p.solve_info().display()`
`p.termination_status` (property)	`p.termination_status()` (Relationship)
`p.objective_value` (property)	`p.objective_value()` (Relationship)
`{{name:float}}` Property shorthand	`{Float:name}` PyRel type syntax

Pattern diversity across templates

Pattern	Templates
IC-heavy (structural checks + solution validation + `.inspect()`)	diet, network_flow, traveling_salesman, production_planning
IC + `solve_info()` (termination IC + Python-side formatted output)	11 templates (hospital_staffing, machine_maintenance, etc.)
IC + `solve_info()` + `.inspect()` (single solve)	ad_spend_allocation, portfolio_balancing, grid_interconnection
`p.verify()` (post-solve constraint checking)	shift_assignment
`solve_info()` in scenario loops (need Python scalars per iteration)	factory_production, supplier_reliability, demand_planning_temporal, sprint_scheduling

Additional changes

Scenario-loop templates now guard against None objective_value on non-OPTIMAL solves
Display-only .to_df() + print replaced with .inspect() in 7 templates

Test plan

All 22 templates pass syntax check (py_compile)
All 22 templates (21 HiGHS + 1 MiniZinc) run successfully against PyRel v1.0.6 on azure-latest
Rebased onto main after PR clean up scenario logic in three prescriptive templates #29 merge; conflict resolution verified
Reviewed by Claude reviewer

🤖 Generated with Claude Code

cafzal

looks good, thanks. we should wait to merge until the new release and update minimum version guidance

chriscoey · 2026-03-19T21:23:37Z

new release is v1.0.6 and was released today

chriscoey · 2026-03-22T19:42:19Z

The preview CI failure is unrelated to this PR's changes — it's a credentials issue.

The workflow checks out relationalai/relationalai-docs using the DOCS_REPO_TOKEN secret, which is returning "Bad credentials". The job retries three times and then fails:

Retrieving the default branch name
Bad credentials - https://docs.github.com/rest
Waiting 15 seconds before trying again
...
##[error]Bad credentials - https://docs.github.com/rest

The DOCS_REPO_TOKEN repo secret likely needs to be rotated by a repo admin. This would affect any PR on this repo, not just this one.

somacdivad · 2026-03-23T14:22:58Z

@chriscoey I'm looking in to the credentials issue

github-actions · 2026-03-23T14:37:21Z

The docs preview for this pull request has been deployed to Vercel!

✅ Preview:	https://relationalai-docs-7g5f4k4lj-relationalai.vercel.app/early-access/pyrel/templates
🔍 Inspect:	https://vercel.com/relationalai/relationalai-docs/FgQrSePWRb2HRNoth2eCLAZmxn6z

somacdivad

Thanks for the updates @chriscoey. I got the preview working again and everything passes there. My one concern is that the README files may need to be updated so that any code changes made here are reflected in any code shown in the READMEs.

Early feedback I got about templates indicated that people want to see some code on the website, but I'm thinking about removing raw code blocks from the READMEs because of the maintenance annoyance.

I haven't settled on a solution yet, though, so I think that for now we just need to keep the README's in sync. There's a prompt in .github/prompts that I've used in the past to easily sync READMEs with an agent.

I'll work on a solution for showing code on the website without having to keep anything in sync so future updates are easier.

Migrate all 22 prescriptive templates to the new result accessor API: - p.display_solve_info() → p.solve_info().display() - p.termination_status/p.objective_value properties → p.termination_status()/p.objective_value() Relationships - Add model.require(p.termination_status() == "OPTIMAL") engine-side ICs to 18 templates - Add structural ICs (num_variables, num_objectives) to network_flow and traveling_salesman - Add p.verify() constraint checking to shift_assignment (MiniZinc) - Replace display-only .to_df()+print with .inspect() in 7 templates - Guard loop-based scenario templates against None objective_value - Fix pre-existing NameError in grid_interconnection (build_formulation) Tested: 20/21 templates pass locally (demand_planning_temporal has pre-existing type issue). Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Update code snippet to match the {Type:name} syntax used in the .py file. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

chriscoey · 2026-03-24T03:18:15Z

Rebased onto main (picking up PR #29's scenario restructuring), resolved conflicts in grid_interconnection, demand_planning_temporal, and sprint_scheduling.

All 22 templates (21 HiGHS + 1 MiniZinc) run successfully against PyRel v1.0.6 after the rebase.

Update code blocks and expected output in 4 template READMEs to reflect the solve_info() API migration and scenario restructuring: - grid_interconnection: scenario loop → single solve with scenario-indexed variables; all code blocks updated - demand_planning_temporal: scenario loop added for planning_end sweep; date filtering/period mapping sections updated - sprint_scheduling: capacity_multiplier scenario loop added; named constraint variables; expected output updated - shift_assignment: constraints extracted to named variables (coverage_ic, workload_ic); p.verify() documented The remaining 15 templates' READMEs don't show any code that was changed (solve_info, .inspect patterns), so no update is needed. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Sections 3-5 referenced `s.solve_for()`, `s.satisfy()`, `s.minimize()` but the code uses `p = Problem(...)`. Fixed to match. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

- sprint_scheduling: fix scenario order (0.5, 0.75, 1.0 not reversed) - machine_maintenance: fix s. → p. in solve/verify code block - shift_assignment: fix s.solve/minimize refs in prose, add solve+verify code block, update infeasibility docs for single-solve approach - demand_planning: fix expected output format (status/cost on separate lines), add planning_end= prefix to summary - grid_interconnection: add explanatory prose to step 5 Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Codex found s.solve_for() and s.minimize() in sections 3-4 that were not covered by the earlier fix. All code blocks now use p. to match the actual code. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

- sprint_scheduling: start_epoch was computed but unused — now used in filter so planning_start actually narrows the backlog - shift_assignment: Shift.capacity loaded from CSV but never used in any constraint — removed dead property to avoid confusion - demand_planning: add comment that SCENARIO_VALUES must be in increasing date order (data accumulates across iterations); clarify safety_stock_weeks is an extension point, not wired in Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

- shift_assignment: restore Shift.capacity and add capacity_ic constraint (max workers per shift from CSV). The property was loaded since the template was created but never enforced. - demand_planning: wire safety_stock_weeks into a constraint requiring inventory >= avg_weekly_demand × safety_stock_weeks per site-SKU pair per week. Previously defined but unused. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

- shifts.csv: increase capacity to 4/5/4 so all scenarios (including coverage_3) are feasible with the new capacity constraint - shift_assignment README: add capacity troubleshooting bullet - demand_planning README: add safety_stock_weeks to INFEASIBLE hint Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Capacity should actually constrain the solution: 3 on Morning/Night means the constraint binds at coverage_3 (exactly 3 assigned). 4 on Afternoon allows slight headroom. Previous 4/5/4 was too loose to demonstrate the constraint's effect. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

The sprint_scheduling code and README described pre-planning_start issues as "backlog items eligible from Sprint 1" — but the start_epoch filter now excludes them. Updated both. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

All .py files use p = Problem(model, ...) but 16 READMEs still showed s.solve_for(), s.satisfy(), s.minimize(), etc. in code blocks. Renamed all to p. for consistency. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Change SCENARIO_VALUES from [0.5, 0.75, 1.0] to [0.35, 0.5, 1.0]. At 0.35x capacity the problem is INFEASIBLE (per-developer limits too tight for 139 story points), demonstrating the scenario sweep's value in finding the breaking point. The code correctly handles infeasibility: prints status, skips results, shows N/A in summary. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

chriscoey · 2026-03-24T21:35:17Z

@somacdivad Good call on the README sync — synced all code blocks across all 21 template READMEs.

What was done:

All READMEs now use p. (matching the .py files) instead of the old s. convention — 16 READMEs were out of sync
Code blocks for grid_interconnection, demand_planning_temporal, sprint_scheduling, and shift_assignment updated to match the scenario restructuring from PR clean up scenario logic in three prescriptive templates #29
Expected output sections updated for the new scenario patterns
Wired two previously-unused parameters into constraints: Shift.capacity (max workers per shift) and safety_stock_weeks (inventory floor)
Fixed start_epoch in sprint_scheduling (was computed but never used in the filter)
Tuned scenario values so the sweep actually demonstrates differentiation (sprint_scheduling now shows an infeasible boundary)

All 21 modified templates verified passing against PyRel v1 on latest.

Re: your longer-term plan for showing code on the website without maintenance burden — happy to help when you have a direction.

somacdivad

Thanks @chriscoey. Looks great!

Resolve conflict in network_flow.py: keep `count` import (used by model.require IC on line 70). Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

- production_planning: rewrite section 2 to describe Scenario-concept single-solve pattern (was incorrectly describing a loop pattern) - supplier_reliability: fix s → p variable name in README snippet - sprint_scheduling: fix scenario values 0.25 → 0.35 to match code Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

chriscoey requested review from jablonskidev and somacdivad as code owners March 19, 2026 07:58

chriscoey requested a review from cafzal March 19, 2026 08:08

cafzal approved these changes Mar 19, 2026

View reviewed changes

github-actions bot temporarily deployed to Preview March 23, 2026 14:32 Inactive

github-actions bot added the deployed label Mar 23, 2026

somacdivad reviewed Mar 23, 2026

View reviewed changes

chriscoey and others added 2 commits March 23, 2026 16:19

Fix demand_planning_temporal README Property syntax

bc76267

Update code snippet to match the {Type:name} syntax used in the .py file. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

chriscoey force-pushed the update-v1-pr730-api-changes branch from f4361ed to bc76267 Compare March 23, 2026 23:46

github-actions bot temporarily deployed to Preview March 23, 2026 23:47 Inactive

chriscoey and others added 11 commits March 24, 2026 12:21

Fix pre-existing s. → p. variable name in demand_planning README

20aa01c

Sections 3-5 referenced `s.solve_for()`, `s.satisfy()`, `s.minimize()` but the code uses `p = Problem(...)`. Fixed to match. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Fix remaining s. → p. refs in machine_maintenance README

d6b7bfd

Codex found s.solve_for() and s.minimize() in sections 3-4 that were not covered by the earlier fix. All code blocks now use p. to match the actual code. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Standardize s. → p. in all template READMEs

d6864f8

All .py files use p = Problem(model, ...) but 16 READMEs still showed s.solve_for(), s.satisfy(), s.minimize(), etc. in code blocks. Renamed all to p. for consistency. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

github-actions bot temporarily deployed to Preview March 24, 2026 21:35 Inactive

chriscoey requested a review from somacdivad March 24, 2026 21:40

somacdivad approved these changes Mar 27, 2026

View reviewed changes

chriscoey and others added 2 commits March 27, 2026 13:20

Merge origin/main into update-v1-pr730-api-changes

e6b0259

Resolve conflict in network_flow.py: keep `count` import (used by model.require IC on line 70). Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Fix ruff import sorting in network_flow.py

b971723

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

github-actions bot temporarily deployed to Preview March 27, 2026 20:22 Inactive

github-actions bot temporarily deployed to Preview March 27, 2026 20:30 Inactive

chriscoey merged commit 5ac323e into main Mar 27, 2026
3 checks passed

chriscoey deleted the update-v1-pr730-api-changes branch March 27, 2026 21:00

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Update v1 templates for PyRel PR #730 API changes#28

Update v1 templates for PyRel PR #730 API changes#28
chriscoey merged 16 commits intomainfrom
update-v1-pr730-api-changes

chriscoey commented Mar 19, 2026 •

edited

Loading

Uh oh!

cafzal left a comment

Uh oh!

chriscoey commented Mar 19, 2026

Uh oh!

chriscoey commented Mar 22, 2026

Uh oh!

somacdivad commented Mar 23, 2026

Uh oh!

github-actions bot commented Mar 23, 2026 •

edited

Loading

Uh oh!

somacdivad left a comment

Uh oh!

chriscoey commented Mar 24, 2026

Uh oh!

chriscoey commented Mar 24, 2026

Uh oh!

somacdivad left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

chriscoey commented Mar 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

API migration (all 22 templates)

Pattern diversity across templates

Additional changes

Test plan

Uh oh!

cafzal left a comment

Choose a reason for hiding this comment

Uh oh!

chriscoey commented Mar 19, 2026

Uh oh!

chriscoey commented Mar 22, 2026

Uh oh!

somacdivad commented Mar 23, 2026

Uh oh!

github-actions bot commented Mar 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

somacdivad left a comment

Choose a reason for hiding this comment

Uh oh!

chriscoey commented Mar 24, 2026

Uh oh!

chriscoey commented Mar 24, 2026

Uh oh!

somacdivad left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

chriscoey commented Mar 19, 2026 •

edited

Loading

github-actions bot commented Mar 23, 2026 •

edited

Loading