www.digitalmars.com         C & C++   DMDScript  

D - Looking for Syntax Help: Structs with no type names

reply Russell Lewis <spamhole-2001-07-16 deming-os.org> writes:
I have some old C++ code I'm porting to D, in which I did a lot of stuff 
like this:

struct foo {
   struct {
     bool element1;
     bool element2;
   } valid;

   bar element1;
   bar element2;
};

So you could access the members "valid.element1" and "valid.element2". 
This is no longer valid code, I have to name the "valid" struct and 
create an instance of the structure.  Is there any more elegant way to 
do it?  I always like the C++ way...creating a struct just to use one 
instance of it seems very hacky to me.
Sep 10 2002
parent reply "Walter" <walter digitalmars.com> writes:
"Russell Lewis" <spamhole-2001-07-16 deming-os.org> wrote in message
news:3D7E352B.50303 deming-os.org...

 I have some old C++ code I'm porting to D, in which I did a lot of stuff
 like this:

 struct foo {
    struct {
      bool element1;
      bool element2;
    } valid;

    bar element1;
    bar element2;
 };

 So you could access the members "valid.element1" and "valid.element2".
 This is no longer valid code, I have to name the "valid" struct and
 create an instance of the structure.  Is there any more elegant way to
 do it?  I always like the C++ way...creating a struct just to use one
 instance of it seems very hacky to me.


Write it like this:

struct foo {
   struct {
     bool xelement1;
     bool xelement2;
   }
   bar element1;
   bar element2;
};

In other words, you can use nested anonymous structs and unions.
Sep 11 2002
next sibling parent reply "Sandor Hojtsy" <hojtsy index.hu> writes:
"Walter" <walter digitalmars.com> wrote in message
news:alpapt$2ifd$3 digitaldaemon.com...

 "Russell Lewis" <spamhole-2001-07-16 deming-os.org> wrote in message
 news:3D7E352B.50303 deming-os.org...

 I have some old C++ code I'm porting to D, in which I did a lot of stuff
 like this:

 struct foo {
    struct {
      bool element1;
      bool element2;
    } valid;

    bar element1;
    bar element2;
 };

 So you could access the members "valid.element1" and "valid.element2".
 This is no longer valid code, I have to name the "valid" struct and
 create an instance of the structure.  Is there any more elegant way to
 do it?  I always like the C++ way...creating a struct just to use one
 instance of it seems very hacky to me.


 Write it like this:

 struct foo {
    struct {
      bool xelement1;
      bool xelement2;
    }


Semicolon missing here?


    bar element1;
    bar element2;
 };


In this case I see no reason to do this. You can just put xelement1 and
xelement2 right into struct foo.
Can you define named types inside a struct? Such as:

struct foo {
   struct valid_str {
     bool element1;
     bool element2;
   } valid;
   bar element1;
   bar element2;
};
Sep 11 2002
parent "Walter" <walter digitalmars.com> writes:
"Sandor Hojtsy" <hojtsy index.hu> wrote in message
news:alpdcb$2l8e$1 digitaldaemon.com...

 In this case I see no reason to do this. You can just put xelement1 and
 xelement2 right into struct foo.


Correct. But when you want to declare a union of structs, this becomes a
useful layout tool.


 Can you define named types inside a struct? Such as:

 struct foo {
    struct valid_str {
      bool element1;
      bool element2;
    } valid;
    bar element1;
    bar element2;
 };


I don't understand what you mean by named type here.
Sep 12 2002
prev sibling parent reply Russell Lewis <spamhole-2001-07-16 deming-os.org> writes:
Walter wrote:

 Write it like this:
 
 struct foo {
    struct {
      bool xelement1;
      bool xelement2;
    }
    bar element1;
    bar element2;
 };
 
 In other words, you can use nested anonymous structs and unions.


Sure, I know that's possible.  But that's no better than just defining 
the x elements in the main struct.  I would really prefer to access the 
elements as "valid.element1".  (Actually, I'd like to make "valid" a 
boolean property of both the element1 and element2 variable...but 
perhaps that's asking too much.

Then again, you could conceivably use this sort of syntax:

struct foo {
     bar element1
       properties {
         bool valid;
       };
     bar element2
       properties {
         bool valid;
       };
}

The compiler would have to store the properties in the struct, but they 
would be accessible using the property syntax;

     foo.element1  // refers to the variable, of type "bar"
     foo.element1.valid // referes to the property
Sep 12 2002
next sibling parent reply "Walter" <walter digitalmars.com> writes:
"Russell Lewis" <spamhole-2001-07-16 deming-os.org> wrote in message
news:3D80CF9E.5070201 deming-os.org...

 Walter wrote:

 Write it like this:

 struct foo {
    struct {
      bool xelement1;
      bool xelement2;
    }
    bar element1;
    bar element2;
 };

 In other words, you can use nested anonymous structs and unions.


 Sure, I know that's possible.  But that's no better than just defining
 the x elements in the main struct.


It becomes useful when doing structs of unions of structs etc.
Sep 12 2002
parent reply Russell Lewis <spamhole-2001-07-16 deming-os.org> writes:
Walter wrote:

 "Russell Lewis" <spamhole-2001-07-16 deming-os.org> wrote in message
 news:3D80CF9E.5070201 deming-os.org...
 

Walter wrote:


Write it like this:

struct foo {
   struct {
     bool xelement1;
     bool xelement2;
   }
   bar element1;
   bar element2;
};

In other words, you can use nested anonymous structs and unions.


Sure, I know that's possible.  But that's no better than just defining
the x elements in the main struct.


 
 
 It becomes useful when doing structs of unions of structs etc.


True, forgot about that.  But what about the rest of it?  Any chance for 
(or value in) the syntax sugar?
Sep 12 2002
parent reply "Walter" <walter digitalmars.com> writes:
"Russell Lewis" <spamhole-2001-07-16 deming-os.org> wrote in message
news:3D80EAC8.5070109 deming-os.org...

 It becomes useful when doing structs of unions of structs etc.


 True, forgot about that.  But what about the rest of it?  Any chance for
 (or value in) the syntax sugar?


I don't understand what the value of it is.
Sep 12 2002
parent reply Russ Lewis <spamhole-2001-07-16 deming-os.org> writes:
Walter wrote:


 "Russell Lewis" <spamhole-2001-07-16 deming-os.org> wrote in message
 news:3D80EAC8.5070109 deming-os.org...

 It becomes useful when doing structs of unions of structs etc.


 True, forgot about that.  But what about the rest of it?  Any chance for
 (or value in) the syntax sugar?


 I don't understand what the value of it is.


The "element1" and "valid.element1" values are tied to each other.  I don't
really want to clutter up the namespace with similar names.  It's particularly
problematic in my circumstance because the D code we're referring to is
automatically generated from user definitions..so what if he defined both an
"element1" and "xelement1" value for his purposes?  Then the user's
"xelement1" conflicts with the automatically generated "xelement1", which is
just a property of the user's "element1".

--
The Villagers are Online! http://villagersonline.com

.[ (the fox.(quick,brown)) jumped.over(the dog.lazy) ]
.[ (a version.of(English).(precise.more)) is(possible) ]
?[ you want.to(help(develop(it))) ]
Sep 13 2002
parent reply "Walter" <walter digitalmars.com> writes:
"Russ Lewis" <spamhole-2001-07-16 deming-os.org> wrote in message
news:3D8201C9.B75E78DD deming-os.org...

 Walter wrote:

 I don't understand what the value of it is.


 The "element1" and "valid.element1" values are tied to each other.  I


don't

 really want to clutter up the namespace with similar names.  It's


particularly

 problematic in my circumstance because the D code we're referring to is
 automatically generated from user definitions..so what if he defined both


an

 "element1" and "xelement1" value for his purposes?  Then the user's
 "xelement1" conflicts with the automatically generated "xelement1", which


is

 just a property of the user's "element1".


Ok, I see. The way to do it is to create a unique prefix for your generated
names, like _Foo_. Disallow the user from using _ in the names he types in,
or disallow leading _.
Sep 13 2002
next sibling parent reply "Walter" <walter digitalmars.com> writes:
I forgot to mention. You can also do things like:

struct Foo
{
    struct Bar { int x; }
    Bar b;
    int x;
}

and refer to b.x.
Sep 13 2002
next sibling parent Russ Lewis <spamhole-2001-07-16 deming-os.org> writes:
Walter wrote:


 I forgot to mention. You can also do things like:

 struct Foo
 {
     struct Bar { int x; }
     Bar b;
     int x;
 }

 and refer to b.x.


Right, that's what I do now, but it's ugly.

--
The Villagers are Online! http://villagersonline.com

.[ (the fox.(quick,brown)) jumped.over(the dog.lazy) ]
.[ (a version.of(English).(precise.more)) is(possible) ]
?[ you want.to(help(develop(it))) ]
Sep 13 2002
prev sibling parent reply "Sandor Hojtsy" <hojtsy index.hu> writes:
"Walter" <walter digitalmars.com> wrote in message
news:alt3lr$235r$1 digitaldaemon.com...

 I forgot to mention. You can also do things like:

 struct Foo
 {
     struct Bar { int x; }
     Bar b;
     int x;
 }

 and refer to b.x.


In C++ you can do:

struct Foo
{
    struct Bar { int x; }b;
    int x;
};

(Walter: That's what I refered to as "named type", as compared to declaring
an *anonymous* struct type inside an other struct.)

and in C++ you can also do:


struct Foo
{
    struct { int x; }b;
    int x;
};

I still don't understand the reason for disallowing these constructs. What's
the problem with them? Too complex for humans?

Sandor
Sep 16 2002
parent "Walter" <walter digitalmars.com> writes:
"Sandor Hojtsy" <hojtsy index.hu> wrote in message
news:am46cp$1gjs$1 digitaldaemon.com...

 I still don't understand the reason for disallowing these constructs.


What's

 the problem with them? Too complex for humans?


The reason is to simplify the parsing.
    struct B { members }
is not a type specifier in D.
Sep 18 2002
prev sibling parent Russ Lewis <spamhole-2001-07-16 deming-os.org> writes:
Walter wrote:


 "Russ Lewis" <spamhole-2001-07-16 deming-os.org> wrote in message
 news:3D8201C9.B75E78DD deming-os.org...

 Walter wrote:

 I don't understand what the value of it is.


 The "element1" and "valid.element1" values are tied to each other.  I


 don't

 really want to clutter up the namespace with similar names.  It's


 particularly

 problematic in my circumstance because the D code we're referring to is
 automatically generated from user definitions..so what if he defined both


 an

 "element1" and "xelement1" value for his purposes?  Then the user's
 "xelement1" conflicts with the automatically generated "xelement1", which


 is

 just a property of the user's "element1".


 Ok, I see. The way to do it is to create a unique prefix for your generated
 names, like _Foo_. Disallow the user from using _ in the names he types in,
 or disallow leading _.


Yeah, I do it like that for some other things, but don't really like it :(


--
The Villagers are Online! http://villagersonline.com

.[ (the fox.(quick,brown)) jumped.over(the dog.lazy) ]
.[ (a version.of(English).(precise.more)) is(possible) ]
?[ you want.to(help(develop(it))) ]
Sep 13 2002
prev sibling parent reply "Sean L. Palmer" <seanpalmer earthlink.net> writes:
Now that sounds cool.

Decorating member variables with little extra functions.  It's be nice to be
able to decorate types that way too.  Especially other classes.

Sean

"Russell Lewis" <spamhole-2001-07-16 deming-os.org> wrote in message
news:3D80CF9E.5070201 deming-os.org...

 Sure, I know that's possible.  But that's no better than just defining
 the x elements in the main struct.  I would really prefer to access the
 elements as "valid.element1".  (Actually, I'd like to make "valid" a
 boolean property of both the element1 and element2 variable...but
 perhaps that's asking too much.

 Then again, you could conceivably use this sort of syntax:

 struct foo {
      bar element1
        properties {
          bool valid;
        };
      bar element2
        properties {
          bool valid;
        };
 }

 The compiler would have to store the properties in the struct, but they
 would be accessible using the property syntax;

      foo.element1  // refers to the variable, of type "bar"
      foo.element1.valid // referes to the property
Sep 13 2002
parent "Sandor Hojtsy" <hojtsy index.hu> writes:
"Sean L. Palmer" <seanpalmer earthlink.net> wrote in message
news:als29h$nng$1 digitaldaemon.com...

 Now that sounds cool.


Indeed.


 Decorating member variables with little extra functions.  It's be nice to


be

 able to decorate types that way too.  Especially other classes.


Well we already have properties for classes. The invention here is that they
behave as having a value of a different (actually built-in) type. And that
is *cast overloading*. If you have cast overloading, then - for example -
you can declare a class whose instances can be directly assigned to and from
integers, and still have your new properties, or boolean fields, or
whatever.

You can already do this in C++.




 "Russell Lewis" <spamhole-2001-07-16 deming-os.org> wrote in message
 news:3D80CF9E.5070201 deming-os.org...

 Sure, I know that's possible.  But that's no better than just defining
 the x elements in the main struct.  I would really prefer to access the
 elements as "valid.element1".  (Actually, I'd like to make "valid" a
 boolean property of both the element1 and element2 variable...but
 perhaps that's asking too much.

 Then again, you could conceivably use this sort of syntax:

 struct foo {
      bar element1
        properties {
          bool valid;
        };
      bar element2
        properties {
          bool valid;
        };
 }

 The compiler would have to store the properties in the struct, but they
 would be accessible using the property syntax;

      foo.element1  // refers to the variable, of type "bar"
      foo.element1.valid // referes to the property
Sep 16 2002