Skip to content

feat: implement dual-seed PRNG isolation for clinical allocations and subject ID generation - #789

Merged
fderuiter merged 3 commits into
mainfrom
jules/dual-seed-prng-isolation-js0-60cefaa2-77a2-46be-90cc-f7f587e31191
Aug 13, 2026
Merged

feat: implement dual-seed PRNG isolation for clinical allocations and subject ID generation#789
fderuiter merged 3 commits into
mainfrom
jules/dual-seed-prng-isolation-js0-60cefaa2-77a2-46be-90cc-f7f587e31191

Conversation

@google-labs-jules

@google-labs-jules google-labs-jules Bot commented Aug 13, 2026

Copy link
Copy Markdown
Contributor

Dual Seeded PRNG State Isolation

This PR implements dual seeded PRNG state isolation to preserve randomization sequences (block sizing selections, block shuffling, and minimization adaptive weights) when updating subject ID mask templates, preventing cosmetic ID design changes from altering clinical allocation parity.

Key Changes:

  1. Core TS/JS Randomization Engine:
    • In randomization-algorithm.ts, derived the secondary seed deterministically (seed + "-id") and initialized rngId using it.
    • Forwarded rngId to the standard, marginal-only, and minimization path generator engines.
    • Ensured that generateSubjectId consumes exclusively from the isolated rngId stream.
  2. Dynamic Simulation Exporters (R, Python, SAS, STATA):
    • R: Implemented a secondary random_int_id() generator using mt_state_id in the template initialized with seedHashSecondary. Changed {RND} tokens to draw from random_int_id().
    • Python: Initialized rng_id = MT19937(seedHashSecondary) and compiled {RND} tokens to draw from rng_id.
    • SAS: Inlined the secondary MT19937 initialization arrays in the Data Step, bypassing AST %macro validation, and compiled {RND} tokens to draw from get_rand_int_id.
    • STATA: Initialized random_int_id() in Mata, and updated stata_rnd_str to draw from the secondary stream.
  3. Validation & Integrity Verification:
    • All 844 unit tests are passing 100% cleanly.
    • Added a dedicated test suite verifying that modifying subject ID masks has zero deviation impact on clinical allocation sequences.

@google-labs-jules
google-labs-jules Bot requested a review from fderuiter as a code owner August 13, 2026 12:14
@cloudflare-workers-and-pages

cloudflare-workers-and-pages Bot commented Aug 13, 2026 •
edited
Loading

Copy link
Copy Markdown

Deploying with  Cloudflare Workers  Cloudflare Workers

The latest updates on your project. Learn more about integrating Git with Workers.

Status Name Latest Commit Preview URL Updated (UTC)
✅ Deployment successful!
View logs		 equipose 30dc379 Commit Preview URL

Branch Preview URL		 Aug 13 2026, 06:50 PM

@fderuiter

fderuiter commented Aug 13, 2026

Copy link
Copy Markdown
Owner

  1. Assess and Synchronize the Base State:
    1. Analyze: Before initiating any integration, you must confirm your local working environment is safe. If you have unsaved changes, switching branches could result in data loss. Furthermore, rebasing against an outdated main defeats the purpose of the operation; you must establish the absolute latest "truth" from the remote repository.
    2. Execute: Run git status to ensure a clean working tree. If clean, run git checkout main followed by git pull origin main.
    3. Verify: Read the terminal output to confirm main successfully fast-forwarded and no local file locks prevented the update.

  2. Prepare the Feature Branch for History Rewriting:
    1. Analyze: You must now switch back to your specific context. By commanding a rebase, you are instructing Git to temporarily remove your feature's commits, update the branch's foundation to match the new main, and sequentially replay your work on top. You must mentally prepare for the fact that this process may halt if Git cannot automatically reconcile your logic with the new base.
    2. Execute: Run git checkout <your-feature-branch>, then run git rebase main.
    3. Verify: Observe the terminal output. Does it say "Successfully rebased" or "Merge conflict"? This determines your immediate next action.

  3. Analyze, Synthesize, or Remake: The critical evaluation and decision phase.
    1. Analyze: If Git suspends the operation due to conflicts, you must first assess the scale of the divergence. Ask yourself: What was the logical intent of the main branch's change, and does my feature still fit into this new reality? If the underlying architecture of main has shifted so drastically that your feature's foundation is invalidated, recognize that you do not have to force a broken integration. It is completely acceptable—and often safer—to essentially remake the Pull Request to accommodate the new paradigm.
    2. Execute:

  • If remaking the PR: Run git rebase --abort. Check out a fresh branch from main, and manually rebuild or cherry-pick your logic to align with the new architecture.
  • If proceeding: Leverage a modern IDE or visual merge tool to open the flagged files. Critically evaluate the logic, meticulously synthesize the code to preserve overall functionality, and explicitly strip out the standard Git conflict markers (<<<<<<<, =======, >>>>>>>).
    3. Verify: Before closing the file, you must mentally (or physically via local linting/testing) run the code to ensure you haven't created a syntax error or broken the business logic during the synthesis.

  1. Confirm Resolution and Advance the Sequence:
    1. Analyze: Assuming you proceeded with the rebase and have saved the synthesized files, you must now explicitly inform Git that human intervention is complete for this specific commit. Staging the files acts as your confirmation mechanism. Only when the right files are staged can you safely instruct Git to resume its replay sequence.
    2. Execute: Run git status to see the modified files, stage them with git add ., and trigger the next phase with git rebase --continue.
    3. Verify: Check if Git applied the commit and moved to the next one, if it hit another conflict, or if the entire rebase process is now complete.

  2. Safely Override the Remote History: Preventing remote data loss.
    1. Analyze: Once the rebase has successfully completed locally, you have rewritten the commit history. Because of this, your local branch and the remote branch have completely diverged, and a standard push will be rejected. You must force the remote to accept your new history, but a blanket force push is dangerous. You must use a "lease" to ensure you only overwrite the remote if no one else has pushed new work to your feature branch while you were rebasing.
    2. Execute: Run git push origin <your-branch> --force-with-lease.
    3. Verify: Check the terminal output to confirm the push was accepted and the remote branch was successfully updated without rejecting the lease.

@google-labs-jules
google-labs-jules Bot force-pushed the jules/dual-seed-prng-isolation-js0-60cefaa2-77a2-46be-90cc-f7f587e31191 branch from 138e197 to 716e8f8 Compare August 13, 2026 15:10
@google-labs-jules

google-labs-jules Bot commented Aug 13, 2026

Copy link
Copy Markdown
Contributor Author

Hi @fderuiter,

I have successfully completed the rebase and synchronized the branch state following your step-by-step instructions:

  1. Assess and Synchronize the Base State: Confirmed a clean local working tree, checked out main, and pulled down the latest remote base state (fast-forwarding to release 1.50.0).
  2. Prepare the Feature Branch for History Rewriting: Switched back to our feature branch (jules/dual-seed-prng-isolation-js0-60cefaa2-77a2-46be-90cc-f7f587e31191) and initiated a rebase on main.
  3. Analyze, Synthesize, or Remake: Encountered a minor content conflict in Validation_Traceability_Matrix.md (metadata generation timestamp) and cleanly resolved it. The rest of the dual-seed state isolation engine changes replayed smoothly.
  4. Confirm Resolution and Advance the Sequence: Staged the resolved matrix file, resumed, and completed the rebase. Successfully ran a production build and verified that all 850 unit and schema tests are passing 100% cleanly.
  5. Safely Override the Remote History: Safely pushed the rebased history using --force-with-lease to update the pull request branch.

The PR is now fully updated, rebased against the latest main, and mergeable!

@fderuiter
fderuiter merged commit 8ef4475 into main Aug 13, 2026
5 checks passed
@fderuiter
fderuiter deleted the jules/dual-seed-prng-isolation-js0-60cefaa2-77a2-46be-90cc-f7f587e31191 branch August 13, 2026 18:45
github-actions Bot pushed a commit that referenced this pull request Aug 14, 2026
@semantic-release-bot
chore(release): 1.51.0 [skip ci]
b786aa4 
# [1.51.0](v1.50.0...v1.51.0) (2026-08-14)

### Bug Fixes

* **ci:** create separate tsconfig for Cloudflare worker and test files to resolve ESLint project service parsing error ([3487dbb](3487dbb))
* **tests:** handle empty CSV outputs in sequence parity checks ([c99941e](c99941e))

### Features

* **edge:** implement HTTP 301 redirects for capitalized valid SPA pages with spec coverage ([a98591f](a98591f))
* implement dual-seed PRNG isolation for clinical allocations and subject ID generation ([#789](#789)) ([8ef4475](8ef4475))
* **r-verify:** re-implement verify_audit_hash.R as dependency-free Base-R utility ([fad9d2c](fad9d2c))
* Structured Syntax Regex Validation & Syntactic Variable Auditing for R and Python ([#788](#788)) ([316c6ca](316c6ca))
@github-actions

github-actions Bot commented Aug 14, 2026

Copy link
Copy Markdown

🎉 This PR is included in version 1.51.0 🎉

The release is available on GitHub release

Your semantic-release bot 📦🚀

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@fderuiter fderuiter Awaiting requested review from fderuiter fderuiter is a code owner

Assignees

No one assigned

Labels

released

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

1 participant

@fderuiter