On Fri, Mar 9, 2018 at 8:19 AM Andrey Tcherepanov <
xnow4fippy...@sneakemail.com> wrote:
> So if this is a "just" a mutex, this whole thing will not be atomic - it
would introduce intermediate (albeit invisible) between "<-" parts. I was
hoping for the "edge collapse" here.
Not sure IIUC, but I should have probably emphasized that the "atomicity"
can be seen over the single send or receive _operation_ only. Within a send
or receive _statement_ the situation is more complex. For instance, in 'ch
<- f()', the RHS is evaluated before the send operation is even attempted.
The send operation may block giving another goroutine a chance, for
example, to call f() again and send to 'ch' before the first caller of
'f()' gets unblocked.
Considering
r := make(chan int, 100)
w := make(chan int, 100)
go func() {
i := 0
for {
ch <- i
i++
}
}
go func() {
for {
w <- <-r
}
}
go func() {
for {
w <- <-r
}
}
We cannot rely on values pulled later from 'w' will to be in order even
though values put into 'r' are guaranteed to be such.
--
-j
--
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.
