Repository navigation
Fix native DXCGC constant initialization and bound parser input - #1476
Open
Qiong Wu (qiowu) (DingmaomaoBJTU) wants to merge 2 commits into
Open
Qiong Wu (qiowu) (DingmaomaoBJTU) wants to merge 2 commits into
Qiong Wu (qiowu) (DingmaomaoBJTU) wants to merge 2 commits into
Conversation
Qiong Wu (qiowu) (DingmaomaoBJTU)
requested a review
from a team
as a code owner
October 16, 2026 07:31
Zhipeng Wang (timenick)
approved these changes
Oct 16, 2026
Zhipeng Wang (timenick)
left a comment
Zhipeng Wang (timenick)
left a comment
Contributor
There was a problem hiding this comment.
Reviewed 066dd71; no actionable findings. Native Release EXE and bridge built successfully. Pytest validation passed all 4 cases, covering existing native regressions, generated FP16/FP32 values and boundaries, and invalid inputs. All CI checks passed. Full-model GPU inference was not verified locally.
🤖 Generated with Codex
Qiong Wu (qiowu) (DingmaomaoBJTU)
changed the title
This branch has not been deployed
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Fix native Raw setup for inline FP32 constants such as Samsung ImageEncoder's 0.35355338454246521, and restore the missing initializer-line length guard in the constant-export fallback.
Single-element FP32 literals are parsed using the classic locale and serialized through the existing upload path. FP16 exact-representation checks remain intact. Nonfinite/out-of-range values and unsupported layouts fail explicitly.
Initializer signatures of 4096 bytes or longer no longer enter MSVC's recursive regex matcher. Oversized unsupported targets fail with a normal diagnostic instead of risking stack overflow; unrelated oversized signatures can be skipped while locating a supported initializer. This guard does not add support for arbitrary large initializer signatures.
Validation: Release EXE and DLL build; CTest passes, including the exact FP32 constant, signed zero, minimum subnormal, invalid literals and a 100,000-character initializer regression. Before the guard-only follow-up, Samsung ImageEncoder passed both Raw and DML-only GPU runs (100 measurements/10 warmups), with CPU cosine approximately 0.999999999992. CLIP/Kokoro have not been rerun with the restored guard, so full-model success is not claimed.