fix Redis bool cache key prefix

[Sahil-4555]

Summary

Fixes RedisCache.GetBool to use the configured Redis key prefix when reading values.

SetBool writes bool values using the prefixed key, but GetBool previously looked up the raw unprefixed key. This meant bool values stored through SetBool would not be found when a Redis prefix was configured.

Changes

• Updated GetBool to read from the same prefixed key format used by SetBool.

[pk910]

Heya @Sahil-4555

Thanks, I went through all the changes. They're technically clean.

Most of this is micro-optimization though (sprintf -> string concat, Replace -> TrimPrefix, %x -> hex.EncodeToString, bytes.Equal, ...) and I don't think it's worth merging just for that. The wins are in the ns range, but these all run inside page handlers that take milliseconds for processing and db queries anyway, so it won't make any measurable difference.

Two things did stand out as actual fixes rather than perf:

• GetBool was missing the key prefix, so it would never match what SetBool writes. There's no caller for it in the app right now so it has no real impact today, but it's still a valid fix.
• The re-check in isSszCompatible after taking the write lock is fine too. Just a note: it's not really a race, the map is already behind the mutex - it just avoids running ValidateType twice when two goroutines hit the write lock at the same time.

Could you pull the GetBool fix out into its own small PR? I'd happily merge that. The rest I'd rather leave out to keep the churn down.

[Sahil-4555]

Heya @Sahil-4555

Thanks, I went through all the changes. They're technically clean.

Most of this is micro-optimization though (sprintf -> string concat, Replace -> TrimPrefix, %x -> hex.EncodeToString, bytes.Equal, ...) and I don't think it's worth merging just for that. The wins are in the ns range, but these all run inside page handlers that take milliseconds for processing and db queries anyway, so it won't make any measurable difference.

Two things did stand out as actual fixes rather than perf:

• GetBool was missing the key prefix, so it would never match what SetBool writes. There's no caller for it in the app right now so it has no real impact today, but it's still a valid fix.
• The re-check in isSszCompatible after taking the write lock is fine too. Just a note: it's not really a race, the map is already behind the mutex - it just avoids running ValidateType twice when two goroutines hit the write lock at the same time.

Could you pull the GetBool fix out into its own small PR? I'd happily merge that. The rest I'd rather leave out to keep the churn down.

Done. let me know if there are any other feedback points.