If you write `_ = s[10]` and len(s)<11, that line will panic. So after that line, the compiler can assume that len(s)>=11 and omit bounds check for smaller indices. That is, if you write
f(s[0]) f(s[1]) … f(s[10]) That will be 11 bounds checks, as depending on what exactly len(s) is, a different line must panic. But if you write _ = s[10] f(s[0]) f(s[1]) … f(s[10]) That will be 1 bounds check, as only one of the lines *can* panic - the first. On Sun, 13 Oct 2024 at 08:24, Nico Braun <rainbowst...@gmail.com> wrote: > Hi, I was under the impression that this, ` _ = s[10] `, happens at > compile time. But that probably doesnt make sense. We dont know the length > of all slices at compile time. > > Then, what is this actually improving? Does the compiler see there will be > a bound check, so it doesnt produce instructions for more bound checks? So > the assembly or bytecode will be smaller. > On Friday 11 October 2024 at 08:34:00 UTC+2 tapi...@gmail.com wrote: > >> On Thursday, October 10, 2024 at 11:51:02 AM UTC+8 Nico Braun wrote: >> >> Hi, recently I became very self aware of potential performance >> implications my code might have. >> >> In go there are a few key concepts, like escape analysis and bound >> checks. If you check the standard library, you can see it was coded with >> micro optimizations. >> >> For example, disabling bound checks in some methods, by doing something >> `like _ = s[10]`. >> >> >> The line is not a no-op. It makes bound checks in advance to remove more >> checks. >> >> >> >> Or trying to use function arguments that they will likely end up on the >> stack and not on the heap, by playing the escape analyzer. I.e. pass buffer >> to io.Reader. Or sometimes, dont pass interface because it is hard for the >> compiler to understand that the interface doesnt escape. >> >> I wonder, if I code like this, is this considered premature optimization. >> Or is more like good habbit? >> >> >> -- > 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/66f9b041-1857-470f-b6c3-ce91a6126c01n%40googlegroups.com > <https://groups.google.com/d/msgid/golang-nuts/66f9b041-1857-470f-b6c3-ce91a6126c01n%40googlegroups.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/CAEkBMfHZyjTb6We1GLZaD2P05K35%2BAXFk6mRUdCcnJycVA2OtQ%40mail.gmail.com.
