www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.learn - How do "pure" member functions work?

reply Sean Eskapp <eatingstaples gmail.com> writes:
Does marking a member function as pure mean that it will return the same
result given the same parameters, or that it will give the same result, given
the same parameters and given the same class/struct members?
Aug 20 2011
parent reply David Nadlinger <see klickverbot.at> writes:
On 8/20/11 5:13 PM, Sean Eskapp wrote:

 Does marking a member function as pure mean that it will return the same
 result given the same parameters, or that it will give the same result, given
 the same parameters and given the same class/struct members?


The second one, the implicit this parameter is just considered a normal 
argument as far as purity is concerned.

David
Aug 20 2011
parent reply Sean Eskapp <eatingstaples gmail.com> writes:
== Quote from David Nadlinger (see klickverbot.at)'s article

 On 8/20/11 5:13 PM, Sean Eskapp wrote:

 Does marking a member function as pure mean that it will return the same
 result given the same parameters, or that it will give the same result, given
 the same parameters and given the same class/struct members?


 The second one, the implicit this parameter is just considered a normal
 argument as far as purity is concerned.
 David


Wait, references and pointers are now valid for pure function arguments?
Aug 20 2011
parent reply Timon Gehr <timon.gehr gmx.ch> writes:
On 08/20/2011 06:24 PM, Sean Eskapp wrote:

 == Quote from David Nadlinger (see klickverbot.at)'s article

 On 8/20/11 5:13 PM, Sean Eskapp wrote:

 Does marking a member function as pure mean that it will return the same
 result given the same parameters, or that it will give the same result, given
 the same parameters and given the same class/struct members?


 The second one, the implicit this parameter is just considered a normal
 argument as far as purity is concerned.
 David


 Wait, references and pointers are now valid for pure function arguments?


There are different forms of pure functions:

weakly pure:
no mutable globals are read or written. the function may however change 
its arguments. weakly pure functions are useful mainly for the 
implementation of functions with stronger purity guarantees.

const pure/strongly pure:
All function arguments are values or const/immutable.

 From the type signature of a function, you can always tell which form 
of pure function it is:

int foo(ref int, int) pure; // weakly pure, could change first argument
int bar(const(int)*, int) pure; // const pure
int qux(immutable(int)*, int) pure; // strongly pure


Weakly pure member functions can therefore change the other members:

struct S{
     int x;
     int foo() pure; // weakly pure, could change x
     int bar() pure const; // const pure, cannot change x
     int qux() pure immutable; // strongly pure, only works with 
immutable instances
}

The rationale of this:
Consider

int baz(int x) pure{
     S s=S(x);
     S.foo();
     return s;
}

This is clearly strongly pure code. If we had no weakly pure, this could 
not be written in that way.
Aug 20 2011
next sibling parent reply Sean Eskapp <eatingstaples gmail.com> writes:
== Quote from Timon Gehr (timon.gehr gmx.ch)'s article

 On 08/20/2011 06:24 PM, Sean Eskapp wrote:

 == Quote from David Nadlinger (see klickverbot.at)'s article

 On 8/20/11 5:13 PM, Sean Eskapp wrote:

 Does marking a member function as pure mean that it will return the same
 result given the same parameters, or that it will give the same result, given
 the same parameters and given the same class/struct members?


 The second one, the implicit this parameter is just considered a normal
 argument as far as purity is concerned.
 David


 Wait, references and pointers are now valid for pure function arguments?


 There are different forms of pure functions:
 weakly pure:
 no mutable globals are read or written. the function may however change
 its arguments. weakly pure functions are useful mainly for the
 implementation of functions with stronger purity guarantees.
 const pure/strongly pure:
 All function arguments are values or const/immutable.
  From the type signature of a function, you can always tell which form
 of pure function it is:
 int foo(ref int, int) pure; // weakly pure, could change first argument
 int bar(const(int)*, int) pure; // const pure
 int qux(immutable(int)*, int) pure; // strongly pure
 Weakly pure member functions can therefore change the other members:
 struct S{
      int x;
      int foo() pure; // weakly pure, could change x
      int bar() pure const; // const pure, cannot change x
      int qux() pure immutable; // strongly pure, only works with
 immutable instances
 }
 The rationale of this:
 Consider
 int baz(int x) pure{
      S s=S(x);
      S.foo();
      return s;
 }
 This is clearly strongly pure code. If we had no weakly pure, this could
 not be written in that way.


Oh, I see, thanks! This isn't documented in the function documentation!
Aug 20 2011
next sibling parent reply bearophile <bearophileHUGS lycos.com> writes:
Sean Eskapp:


 Oh, I see, thanks! This isn't documented in the function documentation!


D purity implementation looks like a simple thing, but it's not simple, it has
several parts that in the last months have be added to the language and
compiler, and we are not done yet, there are few more things to add (like
implicit conversion to immutable of the results of strongly pure functions). It
will need several book pages to document all such necessary design details.

Bye,
bearophile
Aug 20 2011
parent reply Don <nospam nospam.com> writes:
bearophile wrote:

 Sean Eskapp:
 

 Oh, I see, thanks! This isn't documented in the function documentation!


 
 D purity implementation looks like a simple thing, but it's not simple, it has
several parts that in the last months have be added to the language and
compiler, and we are not done yet, there are few more things to add (like
implicit conversion to immutable of the results of strongly pure functions). It
will need several book pages to document all such necessary design details.
 
 Bye,
 bearophile


It is actually very simple: a function marked as 'pure' is not allowed 
to explicitly access any static variables.
Everything else is just compiler optimisation, and the programmer 
shouldn't need to worry about it.
Aug 21 2011
parent reply Timon Gehr <timon.gehr gmx.ch> writes:
On 08/21/2011 09:10 PM, Don wrote:

 bearophile wrote:

 Sean Eskapp:


 Oh, I see, thanks! This isn't documented in the function documentation!


 D purity implementation looks like a simple thing, but it's not
 simple, it has several parts that in the last months have be added to
 the language and compiler, and we are not done yet, there are few more
 things to add (like implicit conversion to immutable of the results of
 strongly pure functions). It will need several book pages to document
 all such necessary design details.

 Bye,
 bearophile


 It is actually very simple: a function marked as 'pure' is not allowed
 to explicitly access any static variables.
 Everything else is just compiler optimisation, and the programmer
 shouldn't need to worry about it.


It can be of value to know that a function is pure as in mathematics if 
it is strongly pure, but can have restricted side-effects if it is 
weakly pure.
Aug 21 2011
parent reply Don <nospam nospam.com> writes:
Timon Gehr wrote:

 On 08/21/2011 09:10 PM, Don wrote:

 bearophile wrote:

 Sean Eskapp:


 Oh, I see, thanks! This isn't documented in the function documentation!


 D purity implementation looks like a simple thing, but it's not
 simple, it has several parts that in the last months have be added to
 the language and compiler, and we are not done yet, there are few more
 things to add (like implicit conversion to immutable of the results of
 strongly pure functions). It will need several book pages to document
 all such necessary design details.

 Bye,
 bearophile


 It is actually very simple: a function marked as 'pure' is not allowed
 to explicitly access any static variables.
 Everything else is just compiler optimisation, and the programmer
 shouldn't need to worry about it.


 
 It can be of value to know that a function is pure as in mathematics if 
 it is strongly pure, but can have restricted side-effects if it is 
 weakly pure.


Well, from the compiler's point of view, it's more complicated than 
that. There are const-pure as well as immutable-pure functions. A 
const-pure function has no side-effects, but cannot be optimised as 
strongly as an immutable-pure function.

BTW: The whole "weak pure"/"strong pure" naming was just something I 
came up with, to convince Walter to relax the purity rules. I'd rather 
those names disappeared, they aren't very helpful.

But the basic point is, that knowing that there are no static variables 
is hugely significant for reasoning about code. The strong pure/weak 
pure distinction is not very interesting (you can distinguish them
using only the function signature).
Aug 22 2011
next sibling parent reply "Simen Kjaeraas" <simen.kjaras gmail.com> writes:
On Mon, 22 Aug 2011 22:19:50 +0200, Don <nospam nospam.com> wrote:


 BTW: The whole "weak pure"/"strong pure" naming was just something I  
 came up with, to convince Walter to relax the purity rules. I'd rather  
 those names disappeared, they aren't very helpful.


The concepts are useful, but better names might be worth it. But what
short word eloquently conveys 'accesses no mutable global state'? :p

What we call strongly pure is what in other languages is simply called
'pure', and that is likely the word that should be used for it.

Weakly pure is a somewhat different beast, and the 'best' solution would
likely be for it to be the default (But as we all know, this would
require changing the language too much. Perhaps in D3...). Noglobal
might be the best we have. My favorite thus far is 'conditionally pure'.
It conveys that the function is pure in certain circumstances, and not
in others. However, it might be somewhat diluted by the addition of pure
inference in newer versions of DMD - that definitely is conditionally
pure.

Const pure is not a concept I'm particularly familiar with. Is this the
special case of calling a conditionally pure function with only
const/immutable parameters, with arguments that are immutable in the
calling context, and that it in those cases can be considered
strongly pure?

-- 
   Simen
Aug 22 2011
parent Don <nospam nospam.com> writes:
Simen Kjaeraas wrote:

 On Mon, 22 Aug 2011 22:19:50 +0200, Don <nospam nospam.com> wrote:
 

 BTW: The whole "weak pure"/"strong pure" naming was just something I 
 came up with, to convince Walter to relax the purity rules. I'd rather 
 those names disappeared, they aren't very helpful.


 
 The concepts are useful, but better names might be worth it. But what
 short word eloquently conveys 'accesses no mutable global state'? :p


 nostatic  ?
stateless ?


 What we call strongly pure is what in other languages is simply called
 'pure', and that is likely the word that should be used for it.




 Weakly pure is a somewhat different beast, and the 'best' solution would
 likely be for it to be the default (But as we all know, this would
 require changing the language too much. Perhaps in D3...). Noglobal
 might be the best we have. My favorite thus far is 'conditionally pure'.
 It conveys that the function is pure in certain circumstances, and not
 in others. However, it might be somewhat diluted by the addition of pure
 inference in newer versions of DMD - that definitely is conditionally
 pure.
 
 Const pure is not a concept I'm particularly familiar with. Is this the
 special case of calling a conditionally pure function with only
 const/immutable parameters, with arguments that are immutable in the
 calling context, and that it in those cases can be considered
 strongly pure?


No, it's where the signature contains const parameters, rather than 
immutable ones.
Two calls to a const-pure function, with the same parameters, may give 
different results. You'd need to do a deep inspection of the parameters,
to see if they changed.

Consider:

x = foo(y);
x = foo(y);
where foo is const-pure.
Perhaps y contains a pointer to x. In that case, foo could depend on x.
Or, there might be a mutable pointer to y somewhere, and x might have an 
opAssign which modifies y.
In each case, the second y is different to the first one.

But, if foo is immutable-pure, it will return the same value both times, 
so one of the calls can be optimized away.
Aug 24 2011
prev sibling parent reply Timon Gehr <timon.gehr gmx.ch> writes:
On 08/22/2011 10:19 PM, Don wrote:

 Timon Gehr wrote:

 On 08/21/2011 09:10 PM, Don wrote:

 bearophile wrote:

 Sean Eskapp:


 Oh, I see, thanks! This isn't documented in the function
 documentation!


 D purity implementation looks like a simple thing, but it's not
 simple, it has several parts that in the last months have be added to
 the language and compiler, and we are not done yet, there are few more
 things to add (like implicit conversion to immutable of the results of
 strongly pure functions). It will need several book pages to document
 all such necessary design details.

 Bye,
 bearophile


 It is actually very simple: a function marked as 'pure' is not allowed
 to explicitly access any static variables.
 Everything else is just compiler optimisation, and the programmer
 shouldn't need to worry about it.


 It can be of value to know that a function is pure as in mathematics
 if it is strongly pure, but can have restricted side-effects if it is
 weakly pure.


 Well, from the compiler's point of view, it's more complicated than
 that. There are const-pure as well as immutable-pure functions. A
 const-pure function has no side-effects, but cannot be optimised as
 strongly as an immutable-pure function.


What significant optimization do immutable-pure functions benefit from 
that const-pure functions cannot? Is it just that the compiler cannot do 
CSE in some cases if there is an impure call in between two const-pure 
calls? (which I think would be rather insignificant)


 BTW: The whole "weak pure"/"strong pure" naming was just something I
 came up with, to convince Walter to relax the purity rules. I'd rather
 those names disappeared, they aren't very helpful.


In some contexts they are in fact useful, otherwise the compiler 
implementation wouldn't have any use for them.


 But the basic point is, that knowing that there are no static variables
 is hugely significant for reasoning about code.  The strong pure/weak
 pure distinction is not very interesting (you can distinguish them
 using only the function signature).


Yes, but saying 'the function is weakly pure' is shorter than saying 
'the pure function signature does include mutable references', just like 
having a flag set to PUREweak is more efficient than examining the 
function signature multiple times.
Aug 22 2011
parent "Jonathan M Davis" <jmdavisProg gmx.com> writes:
On Monday, August 22, 2011 15:57 Timon Gehr wrote:

 On 08/22/2011 10:19 PM, Don wrote:

 Timon Gehr wrote:

 On 08/21/2011 09:10 PM, Don wrote:

 bearophile wrote:

 Sean Eskapp:

 Oh, I see, thanks! This isn't documented in the function
 documentation!


 
 D purity implementation looks like a simple thing, but it's not
 simple, it has several parts that in the last months have be added to
 the language and compiler, and we are not done yet, there are few more
 things to add (like implicit conversion to immutable of the results of
 strongly pure functions). It will need several book pages to document
 all such necessary design details.
 
 Bye,
 bearophile


 
 It is actually very simple: a function marked as 'pure' is not allowed
 to explicitly access any static variables.
 Everything else is just compiler optimisation, and the programmer
 shouldn't need to worry about it.


 
 It can be of value to know that a function is pure as in mathematics
 if it is strongly pure, but can have restricted side-effects if it is
 weakly pure.


 
 Well, from the compiler's point of view, it's more complicated than
 that. There are const-pure as well as immutable-pure functions. A
 const-pure function has no side-effects, but cannot be optimised as
 strongly as an immutable-pure function.


 
 What significant optimization do immutable-pure functions benefit from
 that const-pure functions cannot? Is it just that the compiler cannot do
 CSE in some cases if there is an impure call in between two const-pure
 calls? (which I think would be rather insignificant)


The _only_ time that subsequent calls to a pure function can be optimized out 
is when all of its arguments are immutable or implicitly convertible to 
immutable, otherwise their values could change between calls to the function. 
A function whose parameters are all immutable or implicitly convertible to 
immutable is always going to be able to be optimized that way, because you 
_know_ that it's arguments are immutable or implicitly convertible to 
immutable (they _have_ to be). In cases where a pure function whose parameters 
are const or implicitly convertible to immutable is passed immutable 
arguments, it could be optimized (though that doesn't currently happen), but 
if the arguments aren't immutable, then it can't be, because they could change 
between calls to the function.


 BTW: The whole "weak pure"/"strong pure" naming was just something I
 came up with, to convince Walter to relax the purity rules. I'd rather
 those names disappeared, they aren't very helpful.


 
 In some contexts they are in fact useful, otherwise the compiler
 implementation wouldn't have any use for them.
 

 But the basic point is, that knowing that there are no static variables
 is hugely significant for reasoning about code. The strong pure/weak
 pure distinction is not very interesting (you can distinguish them
 using only the function signature).


 
 Yes, but saying 'the function is weakly pure' is shorter than saying
 'the pure function signature does include mutable references', just like
 having a flag set to PUREweak is more efficient than examining the
 function signature multiple times.


Ultimately, whether pure functions can be optimized or not becomes an 
implementation detail. pure functions can't access mutable global state. 
That's what pure means. The compiler then does what it can to optimize pure 
functions. Weak and strong purity are what determine whether such 
optimizations can take place, but I think that Don is basically saying that 
that should just be an implementation detail and that what should be focused 
on with regards to pure functions is the fact that they can't access mutable 
global state.

- Jonathan M Davis
Aug 22 2011
prev sibling parent Jonathan M Davis <jmdavisProg gmx.com> writes:
On Saturday, August 20, 2011 16:53:51 Sean Eskapp wrote:

 == Quote from Timon Gehr (timon.gehr gmx.ch)'s article
 

 On 08/20/2011 06:24 PM, Sean Eskapp wrote:

 == Quote from David Nadlinger (see klickverbot.at)'s article
 

 On 8/20/11 5:13 PM, Sean Eskapp wrote:

 Does marking a member function as pure mean that it will return
 the same result given the same parameters, or that it will give
 the same result, given the same parameters and given the same
 class/struct members?> >> 


 The second one, the implicit this parameter is just considered a
 normal argument as far as purity is concerned.
 David


 
 Wait, references and pointers are now valid for pure function
 arguments?> 


 There are different forms of pure functions:
 weakly pure:
 no mutable globals are read or written. the function may however change
 its arguments. weakly pure functions are useful mainly for the
 implementation of functions with stronger purity guarantees.
 const pure/strongly pure:
 All function arguments are values or const/immutable.
 
  From the type signature of a function, you can always tell which form
 
 of pure function it is:
 int foo(ref int, int) pure; // weakly pure, could change first argument
 int bar(const(int)*, int) pure; // const pure
 int qux(immutable(int)*, int) pure; // strongly pure
 Weakly pure member functions can therefore change the other members:
 struct S{
 
      int x;
      int foo() pure; // weakly pure, could change x
      int bar() pure const; // const pure, cannot change x
      int qux() pure immutable; // strongly pure, only works with
 
 immutable instances
 }
 The rationale of this:
 Consider
 int baz(int x) pure{
 
      S s=S(x);
      S.foo();
      return s;
 
 }
 This is clearly strongly pure code. If we had no weakly pure, this could
 not be written in that way.


 
 Oh, I see, thanks! This isn't documented in the function documentation!


The documentation takes the approach of saying what functions can be pure but 
not how they can be optimized. Essentially, any function which doesn't access 
any mutable global state and doesn't call any functions which aren't pure can 
be pure. Weak and string purity come in when optimizing. Additional calls to 
weakly pure functions can't be optimized out, because they can alter their 
arguments. Additional calls to strongly pure functions _can_ be optimized out, 
because the compiler can guarantee that its arguments aren't altered.

At present, a strongly pure function is a pure function where all of its 
parameters (including the invisible this parameter) are either immutable or 
implicitly convertible to immutable. But that could be expanded to include 
pure functions whose parameters are all either const or implicitly convertible 
to const in cases where the compiler knows that the function is being passed 
immutable variables. So, right now, David is incorrect in thinking that

int bar(const(int)*, int) pure;

is strongly pure, but in the future, it probably will be in cases where it's 
passed an immutable pointer.

- Jonathan M Davis
Aug 20 2011
prev sibling parent "Steven Schveighoffer" <schveiguy yahoo.com> writes:
On Sat, 20 Aug 2011 12:43:29 -0400, Timon Gehr <timon.gehr gmx.ch> wrote:


 On 08/20/2011 06:24 PM, Sean Eskapp wrote:

 == Quote from David Nadlinger (see klickverbot.at)'s article

 On 8/20/11 5:13 PM, Sean Eskapp wrote:

 Does marking a member function as pure mean that it will return the  
 same
 result given the same parameters, or that it will give the same  
 result, given
 the same parameters and given the same class/struct members?


 The second one, the implicit this parameter is just considered a normal
 argument as far as purity is concerned.
 David


 Wait, references and pointers are now valid for pure function arguments?


 There are different forms of pure functions:

 weakly pure:
 no mutable globals are read or written. the function may however change  
 its arguments. weakly pure functions are useful mainly for the  
 implementation of functions with stronger purity guarantees.


I have to just interject real quick:  One aspect of pure functions that  
usually is forgotten (probably because it's seldom used) is that pure  
functions are not allowed to accept or return shared data.

In other words:

pure int foo(ref shared(int) i); // compiler error.

I think even without the ref, it's a compiler error, although I don't see  
what the point of a shared non-reference type is...

Other than that, you got everything right.

-Steve
Aug 22 2011