On Sat, Mar 13, 2021 at 1:24 AM Matthew Holiday <matthew.holi...@nytimes.com>
wrote:
> I don't think we should change creation; what about having the first
> insert make the map if it's nil?
>
> It seems that would be fairly transparent.
>
I assume you accidentally hit "Reply" instead of "Reply All".
This isn't as simple either, unfortunately:
func F(m map[int]int) {
m[42] = 23
}
func main() {
var m map[int]int
F(m)
fmt.Println(m[42])
}
F might create the map, but the change has to be propagated back somehow -
and all it got is a nil-pointer.
No offense, but again, if it was that simple, it would have already
happened :) No one is arguing that having to explicitly initialize a map
(the same, BTW, goes for channels) is not awkward. But any solution comes
with its own set of problems, be it having to write `*map` everywhere,
introducing call-by-reference semantics, but only for some types, not for
others or significant performance penalties. It's been over a decade and we
didn't find a panacea yet - it's becoming increasingly unlikely that we
find one. And if we where to go through the churn of orchestrating a deep
language change like this, breaking almost every Go program out there, it
really has to be worth it and not just trade one wart for another.
>
> On Fri, Mar 12, 2021 at 5:03 PM 'Axel Wagner' via golang-nuts <
> golang-nuts@googlegroups.com> wrote:
>
>>
>>
>> On Sat, Mar 13, 2021 at 12:01 AM Kevin Chadwick <m8il1i...@gmail.com>
>> wrote:
>>
>>> I do not need to understand the difficulties that exist, though I
>>> struggle to understand the zero bytes property as surely even an empty
>>> string has a 4 or 8 byte pointer.
>>>
>>
>> A string is effectively a
>> struct{
>> data *byte
>> len int
>> }
>> If you zero that, you get a nil pointer and a 0 length, which is a valid
>> representation of an empty string.
>>
>> The same is true for every Go type. A nil pointer is all zero bytes (so
>> pointing at address 0, which is not accessible), maps/funcs/chans are
>> de-facto pointers (we often say "they are pointer-shaped"), so their zero
>> value is also just a pointer to address 0. A nil slice is a
>> struct {
>> data *T
>> len int
>> cap int
>> }
>> which, if zeroed, points at 0 with length/capacity 0. And so on.
>>
>> The zero value of any Go type is just whatever it is represented as in
>> memory, with all bytes set to 0.
>>
>> My original thinking was that either the function call or initialisation
>>> could run make under the covers.
>>
>> From Ian in those threads copied above "While not requiring the make
>>> call".
>>>
>>> Why not require make by calling it, behind the scenes?
>>>
>>
>> But that is exactly the loss of "the zero value is all 0 bytes" we are
>> talking about. It would mean if you write `var m map[int]int` the compiler
>> needs to create instructions to initialize `m` to something that is not
>> zero bytes (a.k.a. "call make behind the scenes"). And if you do `m :=
>> make([]map[int]int, 1e6)` it needs to run those instructions a million
>> times - instead of just getting a block of 0-bytes from the runtime.
>>
>> Of course it's *possible* to abandon this property. Lots of languages
>> (dare I say most languages) don't have it. We'd just prefer not to.
>>
>>
>>>
>>> --
>>> 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/AAB589F5-DA04-4468-A745-77F9177EAF76%40gmail.com
>>> .
>>>
>> --
>> 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/CAEkBMfGUjnqSnBmKZ%3DH6UHWmK%3DSFTigNLnYiBi%3DcT2Km2UzskQ%40mail.gmail.com
>> <https://groups.google.com/d/msgid/golang-nuts/CAEkBMfGUjnqSnBmKZ%3DH6UHWmK%3DSFTigNLnYiBi%3DcT2Km2UzskQ%40mail.gmail.com?utm_medium=email&utm_source=footer>
>> .
>>
>
>
> --
> *Matt Holiday*
> Senior Gopher, Marketing Technologies
>
> 620 Eighth Avenue
>
> New York, NY 10018
>
> matthew.holi...@nytimes.com
>
--
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/CAEkBMfEW2aS7cnUBnPWXqwVu5%3DGp9CqBjTapDRizzeOJ%2Bg2Y1Q%40mail.gmail.com.
