In crypto_rustcrypto.rs there were a few unnecessary .to_vec() calls on byte slices.
This was causing a lot of heap-memory fragmentation which became a problem on a very resource constrained device.
Turned out these allocations are totally avoidable.
Side note: There were two failing tests before (and after) my changes:
In
crypto_rustcrypto.rs
there were a few unnecessary.to_vec()
calls on byte slices.This was causing a lot of heap-memory fragmentation which became a problem on a very resource constrained device. Turned out these allocations are totally avoidable.
Side note: There were two failing tests before (and after) my changes:
data_model::long_reads::test_long_read_subscription_success
data_model::long_reads::test_long_read_success
I haven't had time to look into these but it's unlikely they relate to my changes.