• Jan =?UTF-8?B?SMO2bmln?= (19/19) Nov 07 2020 I have a simple question.
• Paul Backus (6/9) Nov 07 2020 A Tuple is a struct whose members are generated by type sequence
• Jan =?UTF-8?B?SMO2bmln?= (4/6) Nov 08 2020 Is there some recourse, which explains the `alias
• Adam D. Ruppe (14/16) Nov 08 2020 If your object is used in a way that doesn't compile, the
• Jan =?UTF-8?B?SMO2bmln?= (6/22) Nov 08 2020 So it's like inheritance resolved at compile time. It's
• Adam D. Ruppe (2/7) Nov 08 2020 yeah, basically.

Jan =?UTF-8?B?SMO2bmln?= <hrominium gmail.com> writes:

I have a simple question.
How exactly does Tuple work?

In detail, I am interested in expand and opSlice.

For expand, I have found the line: 
https://github.com/dlang/phobos/blob/master/std/typecons.d#L618

How does that work, where is the rest? What does it do?


Similary I can access tuples with `[0]`.
But there is no opSlice.
It is also not specified in the docu: 
https://dlang.org/phobos/std_typecons.html#Tuple


Last but not least, I can use `length`.
How does that work? Where is it?


I probably don't need an explenation, I mainly need to be pointed 
in the right direction.
So instead of sacrificing your time writing an answer, a link to 
the lines I should look at would be fully suficient.

Thanks in advance!

BR,
Jan

Paul Backus <snarwin gmail.com> writes:

On Saturday, 7 November 2020 at 18:02:26 UTC, Jan Hönig wrote:
 I have a simple question.
 How exactly does Tuple work?

 In detail, I am interested in expand and opSlice.

A Tuple is a struct whose members are generated by type sequence 
instantiation:

https://dlang.org/articles/ctarguments.html#type-seq-instantiation

Indexing and slicing are implemented with `alias expand this`, 
which causes `t[i]` to be lowered to `t.expand[i]`.

Jan =?UTF-8?B?SMO2bmln?= <hrominium gmail.com> writes:

On Saturday, 7 November 2020 at 18:31:18 UTC, Paul Backus wrote:
 Indexing and slicing are implemented with `alias expand this`, 
 which causes `t[i]` to be lowered to `t.expand[i]`.

Is there some recourse, which explains the `alias <something> 
this`? I still don't understand what it does. I can't imagine 
what it does to my my class/struct.

Adam D. Ruppe <destructionator gmail.com> writes:

On Sunday, 8 November 2020 at 10:03:46 UTC, Jan Hönig wrote:
 Is there some recourse, which explains the `alias <something> 
 this`?

If your object is used in a way that doesn't compile, the 
compiler will change `obj` to `obj.whatever_alias_this_is` and 
try again.

So say you have

struct S {
     int a;
     alias a this;
}

S obj;

obj += 5;


It will see that obj +=5 doesn't compile on its own, but it has 
alias a this so it changes `obj` to `obj.a` and tries again.

So `obj.a += 5;` is the end result.

Jan =?UTF-8?B?SMO2bmln?= <hrominium gmail.com> writes:

On Sunday, 8 November 2020 at 13:10:33 UTC, Adam D. Ruppe wrote:
 On Sunday, 8 November 2020 at 10:03:46 UTC, Jan Hönig wrote:
 Is there some recourse, which explains the `alias <something> 
 this`?

 If your object is used in a way that doesn't compile, the 
 compiler will change `obj` to `obj.whatever_alias_this_is` and 
 try again.

 So say you have

 struct S {
     int a;
     alias a this;
 }

 S obj;

 obj += 5;


 It will see that obj +=5 doesn't compile on its own, but it has 
 alias a this so it changes `obj` to `obj.a` and tries again.

 So `obj.a += 5;` is the end result.

So it's like inheritance resolved at compile time. It's 
inheritance with virtual member functions without overhead.
I am guessing only one alias works.

And we use this, because struct can't do inheritance and 
interface is abstract.

Adam D. Ruppe <destructionator gmail.com> writes:

On Sunday, 8 November 2020 at 13:57:08 UTC, Jan Hönig wrote:
 So it's like inheritance resolved at compile time. It's 
 inheritance with virtual member functions without overhead.
 I am guessing only one alias works.

 And we use this, because struct can't do inheritance and 
 interface is abstract.

yeah, basically.