Skip to content

Instantly share code, notes, and snippets.

View ryan-williams's full-sized avatar
🚆

Ryan Williams ryan-williams

🚆
View GitHub Profile

[Open-Athena/pulumi#1] Test stack, e2e workflow, and new commands for pulumi.yml

Summary

  • Adds a test Pulumi stack (test/) with lightweight free AWS resources (EC2 key pair, IAM role, CW log group)
  • Adds test.yml caller workflow for manual dispatch against the reusable workflow
  • Adds e2e.yml smoketest: init → preview → up → preview (nop) → destroy → preview (recreate) → stack-rm
  • New commands in pulumi.yml: init, destroy, stack-rm
  • New secrets-provider input for GCP KMS (or other) secrets encryption
  • PULUMI_CONFIG_PASSPHRASE as optional secret input (defaults to empty string for self-managed backends)
  • Run-ID based stack isolation for concurrent e2e runs

[ImageMagick/ImageMagick#8583] Fix double-free in SVG gradientTransform / transform parsing

Summary

Fixes #8582.

In SVGStartElement, the gradientTransform and transform attribute handlers reassign value to tokens[j+1] inside the inner token-parsing loop. After the loop, all tokens (including tokens[j+1]) are freed via DestroyString(). The outer attribute loop then calls DestroyString(value) at line 2524, which double-frees the already-destroyed token string, causing SIGABRT.

  • Use a separate token_value local variable inside each inner loop instead of reassigning value
  • The outer loop's DestroyString(value) now correctly frees the original SVGEscapeString()-allocated string exactly once
  • Add tests/cli-svg.tap regression test

[ImageMagick/ImageMagick#8582] SIGABRT (double-free) converting SVG with gradientTransform on <linearGradient>

Description

magick crashes with a double-free (SIGABRT, exit code 134) when converting any SVG containing a gradientTransform attribute on a <linearGradient> element. Removing the attribute makes conversion succeed. ImageMagick 6 is not affected.

Error output

free(): double free detected in tcache 2
@ryan-williams
ryan-williams / DESCRIPTION.md
Created February 26, 2026 14:23
Draft PR for test-owner/test-repo

Title

Description of the PR...

@ryan-williams
ryan-williams / DESCRIPTION.md
Created February 25, 2026 19:08
Draft PR for test-owner/test-repo

Title

Description of the PR...

@ryan-williams
ryan-williams / DESCRIPTION.md
Created February 24, 2026 23:37
Draft PR for test-owner/test-repo

Title

Description of the PR...

@ryan-williams
ryan-williams / DESCRIPTION.md
Created February 24, 2026 14:56
Draft PR for test-owner/test-repo

Title

Description of the PR...

Summary

Ran pre-commit run -a -v, to get CI passing:

  • Remove unused **kwargs and no-op on_exception method in dataset.py
  • Fix on_test_batch_end signature (add missing _batch param) in lightning.py
  • Use Path.open() instead of Path.open(path, ...) in lightning.py
  • Remove from pathlim import Path typo in lightning.py
  • Wrap long lines in resunet.py

[Quantum-Accelerators/electrai#64] Add GPU CI: e2e tests and benchmarks

Summary

  • Add deterministic e2e training test (tests/e2e_train.py) with platform-specific expected values
  • Add GPU e2e workflow (gpu-e2e.yml) using [ec2-gha] on EC2 g6.xlarge (NVIDIA L4), runs on PRs targeting main
  • Add GPU benchmark workflow (gpu-benchmark.yml) with configurable model size, weekly schedule, and manual dispatch
  • Add WandB logging to benchmark: logs training metrics, model config, dataset version, and instance type
  • Add scripts/s3_sync.py: reusable S3 data sync with size filtering and deterministic dataset hashing
  • Add gen-expected.yml for regenerating expected values on GHA runners (macOS, Ubuntu)
@ryan-williams
ryan-williams / DESCRIPTION.md
Created February 6, 2026 03:45
Draft PR for test-owner/test-repo

Title

Description of the PR...