www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.bugs - [Issue 12170] New: sum(double[]) is not pure

reply d-bugmail puremagic.com writes:
https://d.puremagic.com/issues/show_bug.cgi?id=12170

           Summary: sum(double[]) is not pure
           Product: D
           Version: D2
          Platform: x86
        OS/Version: Windows
            Status: NEW
          Keywords: rejects-valid
          Severity: normal
          Priority: P2
         Component: Phobos
        AssignedTo: nobody puremagic.com
        ReportedBy: bearophile_hugs eml.cc



import std.algorithm;
void main() pure {
    [1.0].sum;
}


dmd 2.065beta3 gives:

test.d(3,10): Error: pure function 'D main' cannot call impure function
'std.algorithm.sum!(double[]).sum'

-- 
Configure issuemail: https://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Feb 15 2014
next sibling parent d-bugmail puremagic.com writes:
https://d.puremagic.com/issues/show_bug.cgi?id=12170


Peter Alexander <peter.alexander.au gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |peter.alexander.au gmail.co
                   |                            |m
         OS/Version|Windows                     |All



06:09:57 PST ---
The problem is this:

"Cyclic functions (i.e. functions that wind up directly or indirectly calling
themselves) are inferred as being impure, throwing, and  system."

http://dlang.org/function.html

sum uses a recursive algorithm for summing floating point ranges for extra
accuracy, so its purity is not inferred. It cannot just be marked as pure
because the range operations it uses may not be pure.

I'm not sure why that cyclic function restriction exists. I think it should be
possible to infer these attributes on recursive functions. I'll have a think.

-- 
Configure issuemail: https://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Feb 15 2014
prev sibling next sibling parent d-bugmail puremagic.com writes:
https://d.puremagic.com/issues/show_bug.cgi?id=12170


bearophile_hugs eml.cc changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|sum(double[]) is not pure   |sum(double[]) is not pure
                   |                            |nothrow



Yes, the same happens with nothrow:


void main() nothrow {
    import std.algorithm: sum;
    [1.0].sum;
}

-- 
Configure issuemail: https://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Feb 15 2014
prev sibling next sibling parent d-bugmail puremagic.com writes:
https://d.puremagic.com/issues/show_bug.cgi?id=12170






 The problem is this:
 
 "Cyclic functions (i.e. functions that wind up directly or indirectly calling
 themselves) are inferred as being impure, throwing, and  system."


Until that compiler limitation is lifted, a solution is to convert the
recursive algorithm of sum() in an iterative one. Because sum() is something
you often want to use in pure nothrow functions.

-- 
Configure issuemail: https://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Feb 15 2014
prev sibling parent d-bugmail puremagic.com writes:
https://d.puremagic.com/issues/show_bug.cgi?id=12170


monarchdodra gmail.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |monarchdodra gmail.com





 Until that compiler limitation is lifted, a solution is to convert the
 recursive algorithm of sum() in an iterative one. Because sum() is something
 you often want to use in pure nothrow functions.


Well, "sum" is specifically implemented to do pair-wise sumation of floats, to
reduce computational error, as opposed to the "dumber" "member by member" sum.

The idea would be to transform the recursive algorithm into an iterative one,
but (AFAIK), this usually requires a stack-type structure.

Either that, or request better inference for cyclic functions.

-- 
Configure issuemail: https://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Feb 27 2014