Use bytes.Buffer instead channels and goroutines #7
Conversation
BenchmarkMakeQueryOld-4 30000 54336 ns/op 2537 B/op 29 allocs/op BenchmarkMakeQueryBuffered-4 200000 6547 ns/op 744 B/op 15 allocs/op
| // Primitive Wrapper Types // | ||
| ///////////////////////////// | ||
| // /////////////////////////// |
|
@gaydin It's nice to see people using it. It's great work to double the speed. May you include your benchmark code as well so that the performance data is reproducible? I'm the original creator of this lib. I don't work for Udacity anymore though. It's up to Udacity folks to decide if they want to merge it. But I am curious how did you find this lib and what's your use cases. |
| "fmt" | ||
| ) | ||
|
|
||
| type argumentValue interface { | ||
| stringChan() <-chan string | ||
| stringChan(buffer *bytes.Buffer) |
There was a problem hiding this comment.
This is fantastic. I wanted to do this when I first created it. But I was too happy about the clever use of channels & goroutine so that I left it there. But I do think an io.Writer alike version is better.
Some suggestions:
- How about
stringWriter() *bytes.Bufferwhich returns an writer instead of consuming one? - Have you think about
io.Writer?
There was a problem hiding this comment.
The function name could be stringWriter so that people don't get confused.
Hello! :)
Thanks for the library
I made a few changes, namely: I got rid of goroutines and channels, which helped greatly increase performance
BenchmarkMakeQueryOld-4 30000 54336 ns/op 2537 B/op 29 allocs/op
BenchmarkMakeQueryBuffered-4 200000 6547 ns/op 744 B/op 15 allocs/op