The documentation <https://golang.org/pkg/os/signal/#Notify> is actually very specific here:
Package signal will not block sending to c: the caller must ensure that c > has sufficient buffer space to keep up with the expected signal rate. *For > a channel used for notification of just one signal value, a buffer of size > 1 is sufficient.* (emphasis mine). This is exactly such a case. We only need one signal value to know the context is cancelled. On Mon, Jul 5, 2021 at 12:55 PM Axel Wagner <axel.wagner...@googlemail.com> wrote: > On Mon, Jul 5, 2021 at 12:41 PM Erman İmer <ermani...@gmail.com> wrote: > >> I think it is not about the context's cancellation. The signal channel >> should have a sufficient buffer size. >> > > 1 is a sufficient buffer size. Note that there is *exactly one* read from > c.ch in > > https://github.com/golang/go/blob/912f0750472dd4f674b69ca1616bfaf377af1805/src/os/signal/signal.go#L287 > Therefore, even if the buffer was larger, any extra buffered signals would > never be read. So the only possible point to having a larger buffer would > be to prevent a writer from blocking on writing to it. > > However, as the documentation you are quoting is pointing out, `Notify` > (the only writer to that channel) writes the signal non-blockingly. > Therefore, there is never any danger of `Notify` getting blocked. The > sentence you are quoting (making sure there is enough buffer size) is there > to assure you have enough buffer space for a signal to get delivered - not > to prevent `Notify` from blocking. > > Lastly, even *if* we needed more buffer space, your assumption that we > need as much buffer space as there are signals is wrong. A single signal > can be caught multiple times. So *if* we needed to be able to handle > multiple signals (we don't - again, there is only one read, which is the > one cancelling the context), we would also take into account that aspect. > And there just is no actually correct size of the buffer, to make sure > *all* signals get delivered - the number of signals that can happen before > a reader is ready is unbounded. > > The code as it is currently is correct. > > > > >> >> Please check the former discussion which determined the Notify function's >> description. >> https://github.com/dominikh/go-staticcheck/issues/30 >> >> Best regards >> >> On Mon, Jul 5, 2021 at 1:28 PM Sean Liao <seankhl...@gmail.com> wrote: >> >>> What problem are you trying to solve? >>> It only makes sense for NotifyContext to receive a single signal since a >>> context can only be cancelled once >>> >>> On Monday, July 5, 2021 at 10:17:49 AM UTC+2 erma...@gmail.com wrote: >>> >>>> Sorry about the broken link, here is the line: >>>> >>>> https://github.com/golang/go/blob/912f0750472dd4f674b69ca1616bfaf377af1805/src/os/signal/signal.go#L284 >>>> >>>> On Monday, July 5, 2021 at 11:15:11 AM UTC+3 erma...@gmail.com wrote: >>>> >>>>> Hello, >>>>> >>>>> I have an improvement idea for the following line in the os/signals >>>>> package. >>>>> >>>>> 284 c.ch = make(chan os.Signal, 1) >>>>> >>>>> Regarding the underlying function's (Notify) description ("the caller >>>>> must ensure that channel has sufficient buffer space to keep up with the >>>>> expected signal rate.") the code can be improved like the following code >>>>> snippet: >>>>> >>>>> cap := len(signals) >>>>> if cap == 0 { >>>>> cap = 1 >>>>> } >>>>> c.ch = make(chan os.Signal, cap) >>>>> >>>>> I wanted to discuss here first before opening my first issue :) >>>>> >>>>> Best regards >>>>> >>>>> >>>>> >>>>> -- >>> You received this message because you are subscribed to a topic in the >>> Google Groups "golang-nuts" group. >>> To unsubscribe from this topic, visit >>> https://groups.google.com/d/topic/golang-nuts/PpUBZ1hxBbM/unsubscribe. >>> To unsubscribe from this group and all its topics, send an email to >>> golang-nuts+unsubscr...@googlegroups.com. >>> To view this discussion on the web visit >>> https://groups.google.com/d/msgid/golang-nuts/c1cd6bd9-5307-4a71-9caa-f13c354185ccn%40googlegroups.com >>> <https://groups.google.com/d/msgid/golang-nuts/c1cd6bd9-5307-4a71-9caa-f13c354185ccn%40googlegroups.com?utm_medium=email&utm_source=footer> >>> . >>> >> >> >> -- >> *Erman İmer* >> >> -- >> 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. >> To view this discussion on the web visit >> https://groups.google.com/d/msgid/golang-nuts/CAAYkT3MhFw3nmx6Fduv0wRiy10uC1un3whkwYN%2BVNbJPvtApfA%40mail.gmail.com >> <https://groups.google.com/d/msgid/golang-nuts/CAAYkT3MhFw3nmx6Fduv0wRiy10uC1un3whkwYN%2BVNbJPvtApfA%40mail.gmail.com?utm_medium=email&utm_source=footer> >> . >> > -- 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. To view this discussion on the web visit https://groups.google.com/d/msgid/golang-nuts/CAEkBMfGrMgVFTAVzZfcruB94rntR6nqWshOC%2Bg%2BTK6FAeaiogw%40mail.gmail.com.
