Confusion between Pointer or No Pointer for Nested array struct

Question

I have read some of the stack overflow question related to "why pointer and why not pointer", but I could not understand much.

So, want to understand based on my example below

I have a list of users and I am finding a hard time to understand which approach is better and faster for the purpose of nesting array struct of users inside another struct.

For an example -

type loc struct {
    Type string
    Point []float64
}

type User struct {
    ID string
    Name string
    Location loc
    ... // 30 more fields
}

Case A - Using pointer for nested array struct

type Games struct {
    ID string
    Owner []*User
    Player []*User
}

// and storing it as

G:= Games{
    ID: "1",
    Owner: []*User{
        &User {
            ID: "2",
            Name: "User A",
        },
    },
}

Case B - Should I use above approach or below approach to

type Games struct {
    ID string
    Owner []User
    Player []User
}

// storing it as

G := Games {
    ID: "1",
    Owner: []User{
        {
            ID: "2",
            Name: "User A",
        },
    },
}

Considering the situation, I do not have to change data for Owner or Player after creating value for Games above and Player can sometimes remain Nil, should I use only User or it's better to use *User

I have following confusion based on above 2 cases

Performance - which one is better & why?
Garbage Collection - which one is better & why?
Any Memory effects?
Any other effects!

Please help me out understanding based on the above example, which method would you have chosen above and why?

I read that and I can understand is there is no point using "pointer" in my use case above. it is perfect to go with Case B. — user3767643, Aug 20 '18 at 03:29
All your question do not have answers: They all depend on other things than pointer vs- non-pointer. If all your method are on *User you probably want to store *Users. For all the "performance" related questions: Measure your actual use case and decide. — Volker, Aug 20 '18 at 03:35
There is no method where the struct is used as pointer or reference. Everything is stored straight away to database after providing value to the struct. That is why, I mentioned to answer based on above situation without assuming anything extra. — user3767643, Aug 20 '18 at 04:11
Okay, so just runtime performance and amount of generated garbage: Measure! Thats why you got tools to do so. There is no sense in guessing or asking (on SO or somewhere other): Measure your specific case! — Volker, Aug 20 '18 at 06:30

thesyncim · Answer 1 · 2018-08-21T00:29:17.097

use option B

but if you want avoid allocation during iterations or change the content, you need to do it like this:

 for i:=range G.Games{
       G.Games[i]....
 }

this way a copy will be created in every iteration

for _,game:=range G.Games{
      game....
 }

this looks like premature optimization. write in the most simple/readable way and if is not fast enough, take a look at this in order to find the bottleneck

Confusion between Pointer or No Pointer for Nested array struct

Please help me out understanding based on the above example, which method would you have chosen above and why?

1 Answers1