www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.learn - gc heap

reply Sam S E <asdf mailinator.com> writes:
What is allocated in the gc heap? Just classes? Dynamic/associative arrays?
Structs? Only things allocated with new? Do built-in types and structs get
deallocated at the end of scope like in C?
Nov 23 2008
parent reply "Jarrett Billingsley" <jarrett.billingsley gmail.com> writes:
On Sun, Nov 23, 2008 at 5:02 PM, Sam S E <asdf mailinator.com> wrote:

 What is allocated in the gc heap? Just classes? Dynamic/associative arrays?
Structs? Only things allocated with new? Do built-in types and structs get
deallocated at the end of scope like in C?


Everything that is a reference type (classes, dynamic arrays, AAs) is
allocated on the heap.

Everything that is a value type (basic types, fixed-size arrays,
structs) are allocated on the stack and are deallocated when the scope
is left.

scope references to classes will delete the class they refer to when
the scope is left.  As a special case, a declaration of the form
"scope x = new ClassType()" will actually allocate the class instance
on the stack instead of on the heap.
Nov 23 2008
parent reply Sam S E <asdf mailinator.com> writes:
Jarrett Billingsley Wrote:


 On Sun, Nov 23, 2008 at 5:02 PM, Sam S E <asdf mailinator.com> wrote:

 What is allocated in the gc heap? Just classes? Dynamic/associative arrays?
Structs? Only things allocated with new? Do built-in types and structs get
deallocated at the end of scope like in C?


 
 Everything that is a reference type (classes, dynamic arrays, AAs) is
 allocated on the heap.
 
 Everything that is a value type (basic types, fixed-size arrays,
 structs) are allocated on the stack and are deallocated when the scope
 is left.
 
 scope references to classes will delete the class they refer to when
 the scope is left.  As a special case, a declaration of the form
 "scope x = new ClassType()" will actually allocate the class instance
 on the stack instead of on the heap.


Thank you; thanks to you I now have a basic understanding of all the features
of D. Now if only Walter could implement all of them :)
--Sam
Nov 23 2008
parent reply "Jarrett Billingsley" <jarrett.billingsley gmail.com> writes:
On Sun, Nov 23, 2008 at 5:35 PM, Sam S E <asdf mailinator.com> wrote:

 Thank you; thanks to you I now have a basic understanding of all the features
of D. Now if only Walter could implement all of them :)
 --Sam


What's that supposed to mean?
Nov 23 2008
parent reply Sam S E <asdf mailinator.com> writes:
Jarrett Billingsley Wrote:


 On Sun, Nov 23, 2008 at 5:35 PM, Sam S E <asdf mailinator.com> wrote:

 Thank you; thanks to you I now have a basic understanding of all the features
of D. Now if only Walter could implement all of them :)
 --Sam


 
 What's that supposed to mean?


There are many useful and interesting features that are planned but
unimplemented.
Nov 23 2008
parent reply "Jarrett Billingsley" <jarrett.billingsley gmail.com> writes:
On Sun, Nov 23, 2008 at 6:48 PM, Sam S E <asdf mailinator.com> wrote:

 Jarrett Billingsley Wrote:


 On Sun, Nov 23, 2008 at 5:35 PM, Sam S E <asdf mailinator.com> wrote:

 Thank you; thanks to you I now have a basic understanding of all the features
of D. Now if only Walter could implement all of them :)
 --Sam


 What's that supposed to mean?


 There are many useful and interesting features that are planned but
unimplemented.


...like what?
Nov 23 2008
parent reply Sam S E <asdf mailinator.com> writes:
Jarrett Billingsley Wrote:


 On Sun, Nov 23, 2008 at 6:48 PM, Sam S E <asdf mailinator.com> wrote:

 Jarrett Billingsley Wrote:


 On Sun, Nov 23, 2008 at 5:35 PM, Sam S E <asdf mailinator.com> wrote:

 Thank you; thanks to you I now have a basic understanding of all the features
of D. Now if only Walter could implement all of them :)
 --Sam


 What's that supposed to mean?


 There are many useful and interesting features that are planned but
unimplemented.


 
 ...like what?


See http://s3.amazonaws.com/dconf2007/WalterAndrei.pdf.
There are some amazing feature listed there. I can't think of anything else I'd
like to see implemented (except for automatic type inference of substructs in
struct literals).

I don't know how to explain those, so I'll give an example:

struct Foo
{
    int a;
    int b;
}

struct Bar
{
    Foo f;
    real asdf;
}

auto def = Bar(Foo(1, 42), 13.7);  //current
Bar def = {{1, 42}, 13.7};            //C-style
                                                 //downside: only possible if
you initialize
                                                 //when you declare
auto def = Bar({1, 42}, 13.7);      //Bar for clarity, Foo obvious
                                        //simpler, especially with many nested
structs

I've gotten pretty far off topic; this probably deserves it's own thread by now.
--Sam
Nov 23 2008
parent reply "Jarrett Billingsley" <jarrett.billingsley gmail.com> writes:
On Sun, Nov 23, 2008 at 10:59 PM, Sam S E <asdf mailinator.com> wrote:

 See http://s3.amazonaws.com/dconf2007/WalterAndrei.pdf.
 There are some amazing feature listed there. I can't think of anything else
I'd like to see implemented (except for automatic type inference of substructs
in struct literals).


Ah, OK.  Yeah, I wonder if a good number of those features will make
it into D anytime soon.  Walter already said macros will be put off
until D3, which were one of the major things I was looking forward to
for D2..  in the mean time, a number of other ideas have come up and
have taken precedence over many things mentioned in the slides.  Some
are already implemented, though.


 I don't know how to explain those, so I'll give an example:

 struct Foo
 {
    int a;
    int b;
 }

 struct Bar
 {
    Foo f;
    real asdf;
 }

 auto def = Bar(Foo(1, 42), 13.7);  //current
 Bar def = {{1, 42}, 13.7};            //C-style
                                                 //downside: only possible if
you initialize
                                                 //when you declare
 auto def = Bar({1, 42}, 13.7);      //Bar for clarity, Foo obvious
                                        //simpler, especially with many nested
structs


Better struct literals would be nice.
Nov 23 2008
parent reply Sam S E <asdf mailinator.com> writes:
Jarrett Billingsley Wrote:


 On Sun, Nov 23, 2008 at 10:59 PM, Sam S E <asdf mailinator.com> wrote:

 See http://s3.amazonaws.com/dconf2007/WalterAndrei.pdf.
 There are some amazing feature listed there. I can't think of anything else
I'd like to see implemented (except for automatic type inference of substructs
in struct literals).


 
 Ah, OK.  Yeah, I wonder if a good number of those features will make
 it into D anytime soon.  Walter already said macros will be put off
 until D3, which were one of the major things I was looking forward to
 for D2..  in the mean time, a number of other ideas have come up and
 have taken precedence over many things mentioned in the slides.  Some
 are already implemented, though.
 

 I don't know how to explain those, so I'll give an example:

 struct Foo
 {
    int a;
    int b;
 }

 struct Bar
 {
    Foo f;
    real asdf;
 }

 auto def = Bar(Foo(1, 42), 13.7);  //current
 Bar def = {{1, 42}, 13.7};            //C-style
                                                 //downside: only possible if
you initialize
                                                 //when you declare
 auto def = Bar({1, 42}, 13.7);      //Bar for clarity, Foo obvious
                                        //simpler, especially with many nested
structs


 
 Better struct literals would be nice.


Better in what way? Do you like my syntax or do you have a better one?
Nov 23 2008
parent reply "Jarrett Billingsley" <jarrett.billingsley gmail.com> writes:
On Mon, Nov 24, 2008 at 12:51 AM, Sam S E <asdf mailinator.com> wrote:

 Better in what way? Do you like my syntax or do you have a better one?


Oh, I was agreeing with you, as well as implying that struct literals
in general could be improved.
Nov 23 2008
parent Sam S E <asdf mailinator.com> writes:
Jarrett Billingsley Wrote:


 On Mon, Nov 24, 2008 at 12:51 AM, Sam S E <asdf mailinator.com> wrote:

 Better in what way? Do you like my syntax or do you have a better one?


 
 Oh, I was agreeing with you, as well as implying that struct literals
 in general could be improved.


You're right, struct literals need to be updated, but I don't think it's
possible to make them any shorter than my suggestion, except maybe a bracket or
two; you need to at least specify the type of the whole thing.
--Sam
Nov 24 2008