www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - Proposal: Static Invariant (C++ concept in D)

reply davidl <davidl 126.com> writes:
I've read a little bit stuff of C++ concept, and find it has some  
similarity with
D's runtime invariant in some abstraction. In some extent , we can imagine  
C++
concept is kind of compile time invariant.

So the following:

class TemplateClass(T)
{
   static invariant()   // reuse keywords, nothing more get introduced
   {
     // prerequisite checking, does T have foo , does T have bar
     Base_Concept_check();  // these emulate the inheritance of concept
     Derived_Concept_check();
   }
}

Thanks for the great discussion & help from h3r3tic, LeoD, Alexander,  
Larsivi
They gave me some comments, and ideas.

Though it might not functionate all possiblities of C++ concept , but it  
brings
us a little bit more powerful way to make sure our templates go all fine.  
Hope
people love this idea :D


DavidL
Sep 17 2007
next sibling parent reply Tom S <h3r3tic remove.mat.uni.torun.pl> writes:
davidl wrote:

 Thanks for the great discussion & help from h3r3tic, LeoD, Alexander, 
 Larsivi
 They gave me some comments, and ideas.



*cough* negative comments *cough*


-- 
Tomasz Stachowiak
http://h3.team0xf.com/
h3/h3r3tic on #D freenode
Sep 17 2007
parent reply Alexander Panek <alexander.panek brainsware.org> writes:
Tom S wrote:

 davidl wrote:

 Thanks for the great discussion & help from h3r3tic, LeoD, Alexander, 
 Larsivi
 They gave me some comments, and ideas.



 *cough* negative comments *cough*
Sep 17 2007
parent reply davidl <davidl 126.com> writes:
在 Mon, 17 Sep 2007 20:45:07 +0800，Alexander Panek  
<alexander.panek brainsware.org> 写道:


 Tom S wrote:

 davidl wrote:

 Thanks for the great discussion & help from h3r3tic, LeoD, Alexander,  
 Larsivi
 They gave me some comments, and ideas.









You didn't raise a valid point. And actually it's you who choose the name.

-- 
使用 Opera 革命性的电子邮件客户程序: http://www.opera.com/mail/
Sep 17 2007
parent reply Alexander Panek <alexander.panek brainsware.org> writes:
davidl wrote:

 You didn't raise a valid point. And actually it's you who choose the name.



I pointed out - as Bill Baxter did here in the newsgroup -, that your 
proposal is rendered useless. "static in" would only add syntactic sugar 
  for a feature that is rather easy to implement in D already.

I pointed out the name, because I thought /if/ there should be a kind of 
compile time constraint, "static variadic" would be the most fitting 
phrase - as opposed to  "static in", which you kept repeating in the 
beginning of our "discussion".
Sep 17 2007
parent reply Alexander Panek <alexander.panek brainsware.org> writes:
Alexander Panek wrote:

 [...] compile time constraint, "static variadic" would be the most fitting
[...]



That should be "static invariant" of course ... !
Sep 17 2007
parent davidl <davidl 126.com> writes:
在 Mon, 17 Sep 2007 22:15:11 +0800，Alexander Panek  
<alexander.panek brainsware.org> 写道:


 Alexander Panek wrote:

 [...] compile time constraint, "static variadic" would be the most  
 fitting [...]



 That should be "static invariant" of course ... !



I replied to bb, and I don't know if that answer good enough for you.

-- 
使用 Opera 革命性的电子邮件客户程序: http://www.opera.com/mail/
Sep 17 2007
prev sibling parent reply Bill Baxter <dnewsgroup billbaxter.com> writes:
davidl wrote:

 I've read a little bit stuff of C++ concept, and find it has some 
 similarity with
 D's runtime invariant in some abstraction. In some extent , we can 
 imagine C++
 concept is kind of compile time invariant.
 
 So the following:
 
 class TemplateClass(T)
 {
   static invariant()   // reuse keywords, nothing more get introduced
   {
     // prerequisite checking, does T have foo , does T have bar
     Base_Concept_check();  // these emulate the inheritance of concept
     Derived_Concept_check();
   }
 }



Can't you just do the same thing now, with slightly different syntax?

For instance Base_Concept_check could be a template mixin with nothing 
but static ifs and other compile-time checks.  Then you just have to say 
"mixin Base_Concept_check;" instead of calling it like a function, but 
same basic thing, no?

--bb
Sep 17 2007
parent reply davidl <davidl 126.com> writes:
在 Mon, 17 Sep 2007 20:55:12 +0800，Bill Baxter  
<dnewsgroup billbaxter.com> 写道:


 davidl wrote:

 I've read a little bit stuff of C++ concept, and find it has some  
 similarity with
 D's runtime invariant in some abstraction. In some extent , we can  
 imagine C++
 concept is kind of compile time invariant.
  So the following:
  class TemplateClass(T)
 {
   static invariant()   // reuse keywords, nothing more get introduced
   {
     // prerequisite checking, does T have foo , does T have bar
     Base_Concept_check();  // these emulate the inheritance of concept
     Derived_Concept_check();
   }
 }



 Can't you just do the same thing now, with slightly different syntax?

 For instance Base_Concept_check could be a template mixin with nothing  
 but static ifs and other compile-time checks.  Then you just have to say  
 "mixin Base_Concept_check;" instead of calling it like a function, but  
 same basic thing, no?

 --bb



it's a little bit different. Because the check means the prerequisite of  
the template,
This is for the whole template scope. not till the last moment of making a  
call. So
a mixin would trigger the "static invariant".

And "static invariant" itself is for prerequisite checking , so it's an  
enforcement for
template design. The template inner scope can only use the stuffs which  
are checked in
the "static invariant", I mean any stuff or members related to type T.  
Also IDE could
provide nice code completion for template writers.

I think it's for elegance. With some prerequisite checking the template  
system would be
nicer.

-- 
使用 Opera 革命性的电子邮件客户程序: http://www.opera.com/mail/
Sep 17 2007
parent Alexander Panek <alexander.panek brainsware.org> writes:
davidl wrote:

 [...] 
 I think it's for elegance. With some prerequisite checking the template 
 system would be
 nicer.



I think this is one of the cases where a feature should be implemented 
in a library, rather than the language specification.
Sep 17 2007