1

Topic: The pointer against the controlled link

Whether there is the basic performance difference between the pointer and the controlled link for significant types? struct A {...} void F (ref A a) {.} void G (A* a) {.} for F and G it is generated identical IL for parameter exceptions a which accepts type ' valuetype A& a ' in F and ' valuetype A* a ' in G.

2

Re: The pointer against the controlled link

Hello, _NN _, you wrote: whether _NN> There is the basic performance difference between the pointer and the controlled link for significant types? It is a little not in a subject, but what for to transfer under the link structure? Same ...

3

Re: The pointer against the controlled link

Hello, Nikolay_Ch, you wrote: N_C> it is a little not in a subject, but what for to transfer under the link structure? Same ... To People it is boring, here and transfer.

4

Re: The pointer against the controlled link

Hello, Nikolay_Ch, you wrote: N_C> it is a little not in a subject, but what for to transfer under the link structure? Same ... To use heavy structures without penalties on copying. It will be useful together with ref readonly, but in the core in infrastructural things of type channels.

5

Re: The pointer against the controlled link

Hello, _NN _, you wrote: whether _NN> There is the basic performance difference between the pointer and the controlled link for significant types? On productivity to you only run the answer give, but I suppose that is not present.  usages are the basic difference: for a case ref it is impossible to transfer null regular means.

6

Re: The pointer against the controlled link

Hello, Sinix, you wrote: S> Hello, Nikolay_Ch, you wrote: N_C>> it is a little not in a subject, but what for to transfer under the link structure? Same ... S> to use heavy structures without penalties on copying. It will be useful together with ref readonly, but in the core in infrastructural things of type channels. If readonly - then I still understand.

7

Re: The pointer against the controlled link

Hello, Nikolay_Ch, you wrote: N_C>>> it is a little not in a subject, but what for to transfer under the link structure? Same ... S>> to use heavy structures without penalties on copying. It will be useful together with ref readonly, but in the core in infrastructural things of type channels. N_C> If readonly - then I still understand. Happens that is necessary as soon as possible, and in that case is there readonly, is not present there readonly - , not to fat. void CrossProduct (ref Vector3f a, ref Vector3f b, ref Vector3f result) works faster, than Vector3f CrossProduct (Vector3f a, Vector3f b) where Vector3f - structure from three float.

8

Re: The pointer against the controlled link

Hello, alexzzzz, you wrote: A> Happens that is necessary as soon as possible, and in that case is there readonly, is not present there readonly - , not to fat. A> void CrossProduct (ref Vector3f a, ref Vector3f b, ref Vector3f result) works faster, than A> Vector3f CrossProduct (Vector3f a, Vector3f b) A> where Vector3f - structure from three float. The third parameter as out can all the same? Though, as a result - probably without a difference.

9

Re: The pointer against the controlled link

Hello, Nikolay_Ch, you wrote: N_C> the third parameter as out Can all the same? Though, as a result - probably without a difference. Yes, as out it is more logical.

10

Re: The pointer against the controlled link

Hello, _NN _, you wrote: whether _NN> There is the basic performance difference between the pointer and the controlled link for significant types? _NN> _NN> struct A {...} _NN> void F (ref A a) {.} _NN> void G (A* a) {.} _NN> it seems to me that the difference on speed is not present. Though when it seems, is better to check . The basic difference that when you accept ref, there for ref precisely something lies the useful; and when you accept the pointer, he can erraticly specify lines knows where.