Dev v0.1.3#2
Merged
Merged
Conversation
sibocw
commented
Jun 19, 2026
sibocw
commented
Collaborator
- More efficient frame drawing: get rid of PNG layer and use raw buffer (no PNG encoding/decoding)
- CI workflows
- Doc site
The example scripts were relocated from src/parallel_animate/examples/ to a top-level examples/ directory, but test_examples.py still imported them as a subpackage, breaking test collection. Import them by module name from the examples/ dir instead. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The producer filled a bounded task queue and only checked worker exit codes after the enqueue loop finished. If a worker died mid-run (e.g. update() raised), it stopped draining the queue; once full, task_queue.put() blocked forever and the error was never surfaced. Real workloads (more frames than queue capacity) would hang indefinitely on any worker exception. Enqueue via _put_or_abort, which puts with a timeout and, whenever the queue stays full, checks worker liveness — tearing down survivors and raising RuntimeError instead of blocking. Add a watchdog-guarded regression test. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Previously each frame was savefig'd to a PNG (zlib compression), written to disk, then read back and decoded by pvio before H.264 encoding — a full compress/disk/decompress cycle per frame purely to hand frames between processes. PNG encoding is CPU-heavy and often dominates render time. Grab pixels straight from the Agg canvas buffer as raw RGB, persist them as uncompressed .npy frames, and stream them lazily to pvio.write_frames_to_video, eliminating the PNG encode/decode entirely. Frames are loaded on demand so peak memory stays bounded. savefig_params now meaningfully supports only dpi (the buffer grab can't honour e.g. bbox_inches); documented accordingly. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Clarify in the Animator docstring that every frame is fully rasterised and that matplotlib blitting is deliberately not used (frames render independently, in separate processes under parallel mode), so speedups come from concurrency, not incremental redraws. Guides users to push frame-invariant work into setup(). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Frame validation only tracked IndexedFrameParams ids, so a positional (enumerate) index colliding with an explicit frame_id — or any repeated index — silently overwrote an earlier frame's file. Validate the final index of every frame (positional or explicit) against a seen-set and raise ValueError on any duplicate. Also wrap the parallel producer loop so any error raised once workers are running (the new validation error, or a worker crash) tears the workers down via _abort_workers rather than leaving them blocked on the queue as orphans. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add guidance to the Animator docstring (and a pointer from the worker) that instance attributes are serialised once per worker in parallel mode, so heavy per-frame data should travel via param_by_frame or be built lazily in setup() rather than stashed on self. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- Extract the IndexedFrameParams unwrap + duplicate-index validation, which was copy-pasted between the serial and parallel render loops, into a shared _resolved_frames generator so the two paths can't drift. - Resolve disable_progress_bar=None to a concrete bool via stderr TTY detection up front, so tqdm and pvio agree: off a TTY both stay quiet, instead of pvio always printing progress into non-interactive logs. Update the affected test. - Replace the stale 'Convert param_by_frame to list' comment (the code only probes len()) and note preload_factor's memory tradeoff in the docstring. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add parallel and serial tests for VideoFrameAnimation, exercising the IndexedFrameParams out-of-order path and generator-based param_by_frame with explicit n_frames, which the other example tests do not cover. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The usage snippet used a {"phase": ...} key, but the referenced
nondeterministic_video_loader.py example passes {"frame": ...}. Match
the example so the docs and code tell the same story.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.