www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.learn - D constness: head & tail

reply drug <drug2004 bk.ru> writes:
what is the state of head&tail constness in D? Where we are and where we 
are moving toward?
Where can I find some info about head and tail constness? I mean forum 
posts, stackoverflow questions, arcticles somewhere and so on.

Thank in advance
Mar 02 2015
next sibling parent reply Jonathan M Davis via Digitalmars-d-learn writes:
On Monday, March 02, 2015 11:21:32 drug via Digitalmars-d-learn wrote:

 what is the state of head&tail constness in D? Where we are and where we
 are moving toward?
 Where can I find some info about head and tail constness? I mean forum
 posts, stackoverflow questions, arcticles somewhere and so on.


There is no head-const in D. const is transitive, so you can never have
anything which is const with anything mutable inside it (other references to
the same data might refer to it as mutable, but never via any const
reference or pointer). However, tail-const is sometimes possible. For
instance, when slicing arrays, you get a tail-const slice (the resulting
array is mutable, but the elements themselves are still const). For
pointers, it's simply a matter of where you put the parens.

const T* foo;
const(T*) foo

both declare a const pointer to a const T, whereas

const(T)* foo

declares a mutable pointer to a const T. But something like

T const(*) foo;

is illegal, becaues it would violate the transivity of const.

For classes, we have std.typecons.Rebindable
(http://dlang.org/phobos/std_typecons.html#Rebindable):

Rebindable!(const T) foo;

Because there is no way to represent a class object separately from its
reference in D, there is no way to indicate that a reference should be
mutable while the class object itself is const, but Rebindable takes care of
that by using a wrapper struct so that the reference can still be const, but
the wrapper itself isn't.

There are mutiple places in the online documentation which discuss const,
e.g.

http://dlang.org/const3.html
http://dlang.org/const-faq.html

There are also several questions on stackoverflow which discuss it. A good
one to look at would be

http://stackoverflow.com/questions/4219600/logical-const-in-d

I also expect that Ali's book discusses it, but I'd have to go digging
through it to find it, and you'd probably be better served reading through
it yourself anyway:

http://ddili.org/ders/d.en/index.html

In any case, AFAIK, where we are with const will be where we will always be
with const in D. It's not something that's generally considered to need more
work. There are some folks who would like to see a way to do Rebindable
built into the language, but that's never happened, and I don't really
expect it to. But other than that, there really hasn't been much discussion
on changing how const works any time recently. So, "where we're moving
toward" is pretty much where we are now.

- Jonathan M Davis
Mar 02 2015
parent reply drug <drug2004 bk.ru> writes:
I'm just trying to study constness in D and periodically fails with it. 
Thanks for answer.
Then I guess that constness in D is in the finished form, hasn't some 
pitfalls, won't be changed significantly in some future and so I just 
need to learn it and understand it?
Mar 02 2015
parent reply Jonathan M Davis via Digitalmars-d-learn writes:
On Monday, March 02, 2015 14:20:45 drug via Digitalmars-d-learn wrote:

 I'm just trying to study constness in D and periodically fails with it.
 Thanks for answer.
 Then I guess that constness in D is in the finished form, hasn't some
 pitfalls, won't be changed significantly in some future and so I just
 need to learn it and understand it?


Pretty much, yeah. const in D is what it's going to be. That involves some
pros and cons in comparison to what C++ has, so in some ways it's going to
be better, and in others, it's going to be frustrating in its restrictions
(e.g. it doesn't work with anything that would require the mutable keyword
in C++), but it's the way that it's going to be in D.

- Jonathan M Davis
Mar 02 2015
parent drug <drug2004 bk.ru> writes:
On 02.03.2015 14:51, Jonathan M Davis via Digitalmars-d-learn wrote:

 On Monday, March 02, 2015 14:20:45 drug via Digitalmars-d-learn wrote:

 I'm just trying to study constness in D and periodically fails with it.
 Thanks for answer.
 Then I guess that constness in D is in the finished form, hasn't some
 pitfalls, won't be changed significantly in some future and so I just
 need to learn it and understand it?


 Pretty much, yeah. const in D is what it's going to be. That involves some
 pros and cons in comparison to what C++ has, so in some ways it's going to
 be better, and in others, it's going to be frustrating in its restrictions
 (e.g. it doesn't work with anything that would require the mutable keyword
 in C++), but it's the way that it's going to be in D.

 - Jonathan M Davis


Yes, const in D is both exciting and frustrating thing. Thanks again.
Mar 02 2015
prev sibling parent reply ketmar <ketmar ketmar.no-ip.org> writes:
or, tl;dr: `const` in D is working as it was designed to work. there are=20
no changes planning for it, and it will not be turned to "c++-like" const.=
Mar 02 2015
next sibling parent Jonathan M Davis via Digitalmars-d-learn writes:
On Monday, March 02, 2015 10:15:00 ketmar via Digitalmars-d-learn wrote:

 or, tl;dr: `const` in D is working as it was designed to work. there are
 no changes planning for it, and it will not be turned to "c++-like" const.


We couldn't even if we wanted to. It would never work with immutable.

- Jonathan M Davis
Mar 02 2015
prev sibling parent reply "Tobias Pankrath" <tobias pankrath.net> writes:
On Monday, 2 March 2015 at 10:15:00 UTC, ketmar wrote:

 or, tl;dr: `const` in D is working as it was designed to work. 
 there are
 no changes planning for it, and it will not be turned to 
 "c++-like" const.


It's working as it's designed to work, although the design is 
somewhat lacking:

https://github.com/D-Programming-Language/phobos/blob/master/std/container/array.d#L505
Mar 02 2015
parent ketmar <ketmar ketmar.no-ip.org> writes:
On Mon, 02 Mar 2015 11:33:00 +0000, Tobias Pankrath wrote:


 It's working as it's designed to work, although the design is somewhat
 lacking:


yeah, i didn't mean that the design is best possible one, but it's set in=20
stone for now.=
Mar 02 2015