Allow overriding files with Plugins, just warn about it

[riebecj]

This fixes a bug where any Plugin would not be able to override any compile-time files.

All Submissions:

• Have you followed the guidelines stated in CONTRIBUTING.md file?
• Have you checked to ensure there aren't any other open Pull Requests for the desired changed?

Type of change

Please delete options that are not relevant.

• Bug fix (non-breaking change which fixes an issue)

New Feature Submission:

• Does your submission pass the tests?
• Have you linted your code locally prior to submission?

Changes To Core Features:

• Have you added an explanation of what your changes do and why you'd like us to include them?
• Have you successfully ran tests with your changes locally?

[greptile-apps]

Greptile Summary

This PR changes how duplicate output paths are resolved during compilation: instead of keeping the first result and silently discarding later ones, it now always uses the last result and emits a warning. The intent is to let plugins override compile-time files.

• compile_results loop now uses last-one-wins semantics with a warning, enabling plugins that contribute to compile_results to override earlier entries.
• The get_static_assets() loop (lines 1253–1263) is unchanged — it still ignores the plugin file entirely when a conflict is detected, so plugins using the static-assets path still cannot override existing files.
• pyi_hashes.json drops 119 package-level .pyi hash entries, retaining only 3 top-level reflex/ hashes (appears to be an unrelated restructuring side-effect).

Confidence Score: 3/5

The core compile_results change is straightforward, but the plugin override goal is only partially met — the static-assets code path still drops plugin files on conflict.

The fix achieves last-one-wins for the compile_results loop, but leaves the get_static_assets() loop unchanged. Plugins that attempt to override via static assets will still have their files silently ignored, which contradicts the stated fix. Merging as-is would leave users confused about why plugin overrides via static assets still do not work.

reflex/compiler/compiler.py — specifically the get_static_assets() loop at lines 1253–1263 that still prevents plugin overrides.

Important Files Changed

Filename	Overview
reflex/compiler/compiler.py	Changed duplicate-path handling in compile_results from "first wins" to "last wins" with a warning, but the parallel get_static_assets() loop still blocks plugin overrides entirely, leaving the stated goal only half-addressed.
pyi_hashes.json	Removes 119 package-level .pyi hash entries (all packages/* paths), keeping only 3 top-level reflex hashes; likely a side-effect of restructuring packages out of the monorepo.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[compile_results loop] --> B{path in output_mapping?}
    B -- Yes --> C[warn: last one will be used]
    C --> D[output_mapping updated - last wins]
    B -- No --> D

    E[plugin get_static_assets loop] --> F{path in output_mapping?}
    F -- Yes --> G[warn: plugin file ignored - BUG NOT FIXED]
    F -- No --> H[output_mapping updated with plugin content]

    I[plugin modify_files_tasks loop] --> J[output_mapping updated via modify fn]

    D --> K[Write all files to Disk]
    H --> K
    G -.->|file dropped| K
    J --> K

Loading

Comments Outside Diff (1)

1. reflex/compiler/compiler.py, line 1253-1263 (link)

   Plugin static assets still can't override existing files

   The PR title says "Allow overriding files with Plugins, just warn about it," but get_static_assets() still silently ignores the plugin file when a conflict is detected (the else branch means the plugin never wins). The first loop now uses "last-one-wins" semantics after a warning, but this second loop keeps the old "first-one-wins/plugin-ignored" behavior, making plugin override via static assets still impossible.

_{Reviews (1): Last reviewed commit: "Allow overriding files with Plugins, jus..." | Re-trigger Greptile}

[codspeed-hq]

Merging this PR will not alter performance

✅ 24 untouched benchmarks
⏩ 2 skipped benchmarks¹

---

_{Comparing riebecj:main (9d56e64) with main (1eed933)}

Footnotes

1. 2 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩