Skip to content

Conversation

@pablogsal
Copy link
Member

@pablogsalpablogsal commented Dec 13, 2025

The live collector's efficiency bar was incorrectly showing 0% success
rate when using filtered modes like --mode exception where no thread had
an active exception. This happened because samples were only counted as
successful when frames were actually processed, conflating "profiler
health" with "filter hit rate".

Samples are now always counted as successful when the profiler can read
from the target process, regardless of whether any frames matched the
current filter. This ensures the efficiency bar accurately reflects
profiler connectivity rather than filter selectivity. The invariant
total_samples == successful_samples + failed_samples is now properly
maintained.

The live collector's efficiency bar was incorrectly showing 0% success rate when using filtered modes like --mode exception where no thread had an active exception. This happened because samples were only counted as successful when frames were actually processed, conflating "profiler health" with "filter hit rate". Samples are now always counted as successful when the profiler can read from the target process, regardless of whether any frames matched the current filter. This ensures the efficiency bar accurately reflects profiler connectivity rather than filter selectivity. The invariant total_samples == successful_samples + failed_samples is now properly maintained.
@pablogsalpablogsal merged commit 52daab1 into python:mainDec 14, 2025
85 of 87 checks passed
@pablogsalpablogsal deleted the tachyon-fix-per branch December 14, 2025 03:31
Sign up for freeto join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant

@pablogsal