mirror of
https://github.com/PrefectHQ/fastmcp.git
synced 2026-10-20 11:51:26 +02:00
2af8a68749#5245 cached build_regex with lru_cache(maxsize=4096). A resource read walks every template in order, so once a process held more than 4096 templates the LRU evicted each pattern before its next use and reads fell back to rebuilding every pattern: 4200 templates read at about 235ms, slower than 4.0.5. The cache is now unbounded; its keys are server-defined template strings, never client input, so it grows only with the templates a server defines. Claude-Session: https://claude.ai/code/session_01KfHgVhbYEhBCC5eSeqGiuG Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com> |
||
|---|---|---|
| .. | ||
| __init__.py | ||
| test_file_resources.py | ||
| test_function_resources.py | ||
| test_resource_security.py | ||
| test_resource_template.py | ||
| test_resource_template_meta.py | ||
| test_resource_template_query_params.py | ||
| test_resources.py | ||
| test_standalone_decorator.py | ||