I saw some Polly code in an application and wondered why it was so complex... I deliberately didnt look at the Polly source code as it would throw my implementation.
I had previously written my own throttle and retry code for Xero so I thought I would try rolling my own version with Timeout for fun.
Example:
var result = await new CrackerBuilder()
.Throttle(callLimit: 100, period: TimeSpan.FromSeconds(60))
.Timeout(TimeSpan.FromSeconds(1))
.Retry(retryAttempts: 3)
.WithDelayBetweenRetries(attempt => TimeSpan.FromSeconds(Math.Pow(2, attempt)))
.UnessException<CustomException>()
.UnessException<HttpException>(x => x.ErrorCode = 404)
.ExecuteAsync(job.ExecuteAsync);
TODO:
- Timeout applies to the entire call stack, it should only apply for the calls after it on the bulder. e.g.
// Timeout after 1 second, attempt up to 3 times, .
var result = await new CrackerBuilder()
.Timeout(TimeSpan.FromSeconds(1))
.Retry(retryAttempts: 3)
.ExecuteAsync(job.ExecuteAsync);
// vrs
// Attempt 3 times, timeout after 1 second on each try.
var result = await new CrackerBuilder()
.Retry(retryAttempts: 3)
.Timeout(TimeSpan.FromSeconds(1))
.ExecuteAsync(job.ExecuteAsync);