www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.bugs - [Issue 9339] New: std.random.uniform!Enum should return random enum member

reply d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339

           Summary: std.random.uniform!Enum should return random enum
                    member
           Product: D
           Version: D2
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: enhancement
          Priority: P2
         Component: Phobos
        AssignedTo: nobody puremagic.com
        ReportedBy: hsteoh quickfur.ath.cx



Currently, std.random.uniform does not respect enum bounds:

import std.random;
import std.stdio;

enum Fruit {
        Apple = 14,
        Orange = 27,
        Pear = 36,
        Mango = 47
}

void main() {
        writefln("%d", Fruit.min);
        writefln("%d", Fruit.max);

        writeln(uniform!Fruit());
}


Typical output:

14
47
cast(Fruit)-2046817621


It should at the very least respect the enum range defined by the enum's .min
and .max properties (which in this case are 14 and 47, respectively).

Ideally, it should only return one of the four possible values of Fruit, not
anything outside the range of .min and .max, and not anything in between the
four possible values.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 17 2013
next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339


Andrej Mitrovic <andrej.mitrovich gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |andrej.mitrovich gmail.com



16:31:49 PST ---
Would this suffice?

auto uniform(T)()
if (is(T == enum) && isIntegral!T || isSomeChar!T)
{
    enum arr = [EnumMembers!T];
    return randomSample(arr, 1);
}

(You would have to add a !is(T == enum) in the other template).

Example:

import std.random;
import std.stdio;
import std.traits;

enum Fruit
{
    Apple  = 14,
    Orange = 27,
    Pear   = 36,
    Mango  = 47
}

auto uniform(T)()
if (is(T == enum) && isIntegral!T || isSomeChar!T)
{
    enum arr = [EnumMembers!T];
    return randomSample(arr, 1);
}

void main()
{
    foreach (i; 0 .. 10)
        writeln(uniform!Fruit());
}

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 17 2013
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339




Yeah, that will do. Except that the "enum arr = [EnumMembers!E];" line may run
into issue 6057. :)

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 17 2013
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339


bearophile_hugs eml.cc changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |bearophile_hugs eml.cc






 auto uniform(T)()
 if (is(T == enum) && isIntegral!T || isSomeChar!T)
 {
     enum arr = [EnumMembers!T];
     return randomSample(arr, 1);
 }



I think this is more efficient:


T uniform(T)()
if (is(T == enum) && isIntegral!T || isSomeChar!T)
{
    static immutable T[EnumMembers!T.length] members = [EnumMembers!T];
    return members[std.random.uniform(0, members.length)];
}

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 17 2013
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339







     enum arr = [EnumMembers!T];


Be very careful with enum arrays. They are very inefficient.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 17 2013
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339





[...]

 I think this is more efficient:
 
 
 T uniform(T)()
 if (is(T == enum) && isIntegral!T || isSomeChar!T)
 {
     static immutable T[EnumMembers!T.length] members = [EnumMembers!T];
     return members[std.random.uniform(0, members.length)];
 }


You're right, we want the array to be statically initialized.

Does enum arr = [...] cause the code to create the array every time the
function is called?

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 17 2013
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339




17:01:43 PST ---



 

     enum arr = [EnumMembers!T];


 
 Be very careful with enum arrays. They are very inefficient.


Well, the compiler is very inefficient, static will do.


 T uniform(T)()
 if (is(T == enum) && isIntegral!T || isSomeChar!T)
 {
     static immutable T[EnumMembers!T.length] members = [EnumMembers!T];
     return members[std.random.uniform(0, members.length)];
 }


That's not doing what was requested.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 17 2013
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339




17:08:20 PST ---
As much as I'd love to make a pull for this I already know I'm going to run
into Issue 6057 (which has a pull but needs a review).

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 17 2013
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339




17:29:52 PST ---


 As much as I'd love to make a pull for this I already know I'm going to run
 into Issue 6057 (which has a pull but needs a review).




-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 17 2013
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339




If you write "static arr = [EnumMembers!T];", you should be able to evade issue
6057.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 17 2013
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339




17:43:00 PST ---


 If you write "static arr = [EnumMembers!T];", you should be able to evade issue
 6057.


The issue is with unittests.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 17 2013
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339




I don't understand. If you use that line in uniform(), and it works, then
unittests shouldn't have any problems either, no?

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 17 2013
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339




21:21:50 PST ---


 I don't understand. If you use that line in uniform(), and it works, then
 unittests shouldn't have any problems either, no?


The problem is the enum has to be hidden in a unittest block like so:

version(unittest)
{
   enum TestEnum { ... }
}

unittest
{
    foreach (_; 0 .. 100)
        assert(uniform!TestEnum() == ...);
}

And this causes linking problems due to Issue 6057.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 17 2013
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339




Oh? This code compiles & links just fine:

import std.random;
import std.traits;

E randomPick(E)() if (is(E == enum)) {
        static members = [ EnumMembers!E ];
        return members[uniform(0, EnumMembers!E.length)];
}

void main() {
}

unittest {
        enum Fruit { Apple = 12, Mango = 29, Pear = 72 };
        foreach (_; 0 .. 100) {
                auto f = randomPick!Fruit();
                assert(f == Fruit.Apple || f == Fruit.Mango || f ==
Fruit.Pear);
        }
}

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 17 2013
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339







 T uniform(T)()
 if (is(T == enum) && isIntegral!T || isSomeChar!T)
 {
     static immutable T[EnumMembers!T.length] members = [EnumMembers!T];
     return members[std.random.uniform(0, members.length)];
 }


 
 That's not doing what was requested.


Then I don't understand. This ER asks for that function overload to return a
"random enum member". Isn't members[std.random.uniform(0, members.length)] a
random enum member?

And beside what the OP is asking, uniform() returns single random values of a
type. Isn't this what I am doing there?

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 18 2013
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339




10:30:47 PST ---



 

 T uniform(T)()
 if (is(T == enum) && isIntegral!T || isSomeChar!T)
 {
     static immutable T[EnumMembers!T.length] members = [EnumMembers!T];
     return members[std.random.uniform(0, members.length)];
 }


 
 That's not doing what was requested.


 
 Then I don't understand. 


I've misread the last statement (it was late) as:

return std.random.uniform(0, members.length);

instead of:

return members[std.random.uniform(0, members.length)];

Apologies.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 18 2013
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339




17:01:19 PST ---
https://github.com/D-Programming-Language/phobos/pull/1086

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 21 2013
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339






 https://github.com/D-Programming-Language/phobos/pull/1086


I have seen that of this line you have dropped both the immutable and fixed
sized array, can you explain why a dynamic array is better than a fixed array
in the static segment?

static immutable T[EnumMembers!T.length] members = [EnumMembers!T];

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 21 2013
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339




17:21:07 PST ---




 https://github.com/D-Programming-Language/phobos/pull/1086


 
 I have seen that of this line you have dropped both the immutable and fixed
 sized array, can you explain why a dynamic array is better than a fixed array
 in the static segment?
 
 static immutable T[EnumMembers!T.length] members = [EnumMembers!T];


It was a mistake, I've updated the code now.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jan 21 2013
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339




Commits pushed to master at https://github.com/D-Programming-Language/phobos

https://github.com/D-Programming-Language/phobos/commit/c95100ba78cc137ee4af37d59e3b3dfb704a4832
Fixes Issue 9339 - Uniform for enums.

https://github.com/D-Programming-Language/phobos/commit/6a3ffa5e136d22b31529e6a3688cb8ce3a5508a0


Issue 9339 - std.random.uniform for enums

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Feb 12 2013
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339


Andrej Mitrovic <andrej.mitrovich gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED


-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Feb 12 2013
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339




One test case to try:


import std.bigint, std.random;

enum Foo : BigInt {
    zero = BigInt(0),
    one =  BigInt(1)
}

void main() {
    auto x = uniform!Foo();
}

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Feb 12 2013
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339




17:38:45 PST ---


 One test case to try:
 
 
 import std.bigint, std.random;
 
 enum Foo : BigInt {
     zero = BigInt(0),
     one =  BigInt(1)
 }
 
 void main() {
     auto x = uniform!Foo();
 }


It works. It should actually work with any enum base types, since the only
requirement is that the type is an enum.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Feb 12 2013
prev sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=9339




Commit pushed to staging at https://github.com/D-Programming-Language/phobos

https://github.com/D-Programming-Language/phobos/commit/128eaa901ccc0b011773e269e7f2adcd8e0d03b8
Fixes Issue 9339 - Uniform for enums.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Feb 13 2013