Does it seem reasonable that the demangle() function should work for the .mangleof property?


private import std.stdio;
private import std.demangle;

void main() {
	class A {
	}
	writefln(A.mangleof);
	writefln(demangle(A.mangleof));
}

This code prints

C_Dmain1A
C_Dmain1A

It would be nice if it printed

C_Dmain1A
class main.A


Does the D language even have the concept of a fully qualified class name? The property .classinfo.name is just the simple name of a class.

The AssertError also just prints the simple file name rather than the source file path. This makes the error less useful when having modules with the same name in different packages.

Thanks,
  Bradley

Bradley Smith wrote:
> 
> Does it seem reasonable that the demangle() function should work for the .mangleof property?

It seems very reasonable to me. I have implemented compile-time demangle metafunctions:

static assert( prettynameof!(A)    == "class main.A" );
static assert( qualifiednameof!(A) == "main.A" );
static assert( symbolnameof!(A)    == "A" );

When doing this, I found some mistakes in the name mangling docs, and I'm pretty sure that my code is much more comprehensive than std.demangle.

A really old version of the code is in dsource.ddl.meta, but it was ugly, and improvements to the template system have broken it. I'll post my new version soon (was working on it today, still fails in a few cases).

> 
> 
> private import std.stdio;
> private import std.demangle;
> 
> void main() {
>     class A {
>     }
>     writefln(A.mangleof);
>     writefln(demangle(A.mangleof));
> }
> 
> This code prints
> 
> C_Dmain1A
> C_Dmain1A
> 
> It would be nice if it printed
> 
> C_Dmain1A
> class main.A

> 
> 
> Does the D language even have the concept of a fully qualified class name? The property .classinfo.name is just the simple name of a class.
> 
> The AssertError also just prints the simple file name rather than the source file path. This makes the error less useful when having modules with the same name in different packages.
> 
> Thanks,
>   Bradley

Don Clugston wrote:
> Bradley Smith wrote:
>>
>> Does it seem reasonable that the demangle() function should work for the .mangleof property?
> 
> It seems very reasonable to me. I have implemented compile-time demangle metafunctions:
> 
> static assert( prettynameof!(A)    == "class main.A" );
> static assert( qualifiednameof!(A) == "main.A" );
> static assert( symbolnameof!(A)    == "A" );
> 
> When doing this, I found some mistakes in the name mangling docs, and I'm pretty sure that my code is much more comprehensive than std.demangle.
> 
> A really old version of the code is in dsource.ddl.meta, but it was ugly, and improvements to the template system have broken it. I'll post my new version soon (was working on it today, still fails in a few cases).
> 


I eagerly await your improvements. :)

-JJR

John Reimer wrote:
> Don Clugston wrote:
>> Bradley Smith wrote:
>>>
>>> Does it seem reasonable that the demangle() function should work for the .mangleof property?
>>
>> It seems very reasonable to me. I have implemented compile-time demangle metafunctions:
>>
>> static assert( prettynameof!(A)    == "class main.A" );
>> static assert( qualifiednameof!(A) == "main.A" );
>> static assert( symbolnameof!(A)    == "A" );
>>
>> When doing this, I found some mistakes in the name mangling docs, and I'm pretty sure that my code is much more comprehensive than std.demangle.
>>
>> A really old version of the code is in dsource.ddl.meta, but it was ugly, and improvements to the template system have broken it. I'll post my new version soon (was working on it today, still fails in a few cases).
>>
> 
> 
> I eagerly await your improvements. :)
> 
> -JJR

It's done.
I've also added prettytypeof!(T) which converts any type into a string.

auto x = 5.0;
static assert( prettytypeof!( x ) == "double");

These functions works with just about anything I've thrown at them. Try to break it!
Enjoy!

Don Clugston wrote:
> It's done.
> I've also added prettytypeof!(T) which converts any type into a string.
> 
> auto x = 5.0;
> static assert( prettytypeof!( x ) == "double");

Cool :)


> These functions works with just about anything I've thrown at them. Try to break it!
> Enjoy!

Sure thing !

http://158.75.59.9/~h3/code/nameofBreak.rar


-- 
Tomasz Stachowiak  /+ a.k.a. h3r3tic +/

Tom S wrote:
> Don Clugston wrote:
>> It's done.
>> I've also added prettytypeof!(T) which converts any type into a string.
>>
>> auto x = 5.0;
>> static assert( prettytypeof!( x ) == "double");
> 
> Cool :)
> 
> 
>> These functions works with just about anything I've thrown at them. Try to break it!
>> Enjoy!
> 
> Sure thing !
> 
> http://158.75.59.9/~h3/code/nameofBreak.rar

That was quick! Hopefully I'll post a fix tonight.
I notice you're using the 'meta' namespace for your Tuple code. How about joining forces and making a compelling D metaprogramming library?
Previously, my 'meta' library was just proof-of-concept stuff, but it's time to turn it into something serious.
Oskar's done a lot of fantastic stuff too, although his seems to be more relevant at run time.

Don Clugston wrote:
> I notice you're using the 'meta' namespace for your Tuple code. How about joining forces and making a compelling D metaprogramming library?
> Previously, my 'meta' library was just proof-of-concept stuff, but it's time to turn it into something serious.
> Oskar's done a lot of fantastic stuff too, although his seems to be more relevant at run time.

I'd be more than eager to cocreate such a library.
There's a D lib in the works that could use a 'meta' component... :>


-- 
Tomasz Stachowiak  /+ a.k.a. h3r3tic +/

Tom S wrote:
> Don Clugston wrote:
>> I notice you're using the 'meta' namespace for your Tuple code. How about joining forces and making a compelling D metaprogramming library?
>> Previously, my 'meta' library was just proof-of-concept stuff, but it's time to turn it into something serious.
>> Oskar's done a lot of fantastic stuff too, although his seems to be more relevant at run time.
> 
> I'd be more than eager to cocreate such a library.
> There's a D lib in the works that could use a 'meta' component... :>

I'm in. If you make arrangements to set up the meta directory, I'll start contributing. Let's get this ship afloat.

Don Clugston wrote:
> Tom S wrote:
>> Don Clugston wrote:
>>> It's done.
>>> I've also added prettytypeof!(T) which converts any type into a string.
>>>
>>> auto x = 5.0;
>>> static assert( prettytypeof!( x ) == "double");
>>
>> Cool :)
>>
>>
>>> These functions works with just about anything I've thrown at them. Try to break it!
>>> Enjoy!
>>
>> Sure thing !
>>
>> http://158.75.59.9/~h3/code/nameofBreak.rar
> 
> That was quick! Hopefully I'll post a fix tonight.

I've fixed it. Try to break it again!

Don Clugston wrote:
> I've fixed it. Try to break it again!

http://158.75.59.9/~h3/code/kknd.rar

"Shall we play a game ?" :D


-- 
Tomasz Stachowiak  /+ a.k.a. h3r3tic +/