Hi Team,
I'm reaching out regarding a performance optimization opportunity we've identified in our Rust application.
Current Situation: Our application uses PartiqlCatalog to evaluate PartiQL expressions. During recent profiling analysis, we discovered that repeatedly initializing PartiqlCatalog instances is consuming significant CPU cycles and impacting overall application performance.
Challenge: We explored initializing PartiqlCatalog once during application startup to avoid repeated initialization overhead. However, we found that the current implementation doesn't support sharing PartiqlCatalog objects across threads, which is essential for our multi-threaded architecture.
Request: Could we please explore making PartiqlCatalog thread-safe (implementing Send and Sync traits) to enable safe sharing across threads? This enhancement would allow us to:
Initialize the catalog once during startup
Reuse the same instance across multiple threads
Significantly reduce CPU overhead from repeated initializations
Thank you
Hi Team,
I'm reaching out regarding a performance optimization opportunity we've identified in our Rust application.
Current Situation: Our application uses PartiqlCatalog to evaluate PartiQL expressions. During recent profiling analysis, we discovered that repeatedly initializing PartiqlCatalog instances is consuming significant CPU cycles and impacting overall application performance.
Challenge: We explored initializing PartiqlCatalog once during application startup to avoid repeated initialization overhead. However, we found that the current implementation doesn't support sharing PartiqlCatalog objects across threads, which is essential for our multi-threaded architecture.
Request: Could we please explore making PartiqlCatalog thread-safe (implementing Send and Sync traits) to enable safe sharing across threads? This enhancement would allow us to:
Initialize the catalog once during startup
Reuse the same instance across multiple threads
Significantly reduce CPU overhead from repeated initializations
Thank you