[go-nuts] How can race timing inject io.EOF into bufio.NewReader(os.Stdin) ?

[Everton Marques]

I am debugging an issue that randomly makes the code below to hit io.EOF 
instead of reporting first byte 'u' from "ugh":

$ echo "ugh" | eraseme ;# this randomly reports EOF instead of first byte 
'u' from "ugh"

$ more eraseme/main.go
// (snip)
func main() {
        input := bufio.NewReader(os.Stdin)
        b, err := input.ReadByte()
        log.Printf("byte='%c' error=%v", b, err)
        if err != nil {
                os.Exit(1)
        }
}

I suspect the offending code is a variable initialization from an imported 
package:

var Result = newResult() // in imported package

I can suppress the race by not importing that package or by turning 
newResult() into a no-op function.

However, all newResult() does it to allocate/initialize variables and 
log.Printf() messages.

I fail to see how newResult() could directly affect reading from os.Stdin.

Question: Can anyone figure out how initialization timing (?) would insert 
io.EOF in the stream created by bufio.NewReader(os.Stdin) ?

System is: go version go1.12 linux/amd64

The code does NOT launch any goroutine.

Thanks,
Everton

-- 
You received this message because you are subscribed to the Google Groups 
"golang-nuts" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to golang-nuts+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.