www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.learn - AA keys

reply bearophile <bearophileHUGS lycos.com> writes:
Can you explain this to me?
Is this yet another corner case (or bug?)?

import std.stdio: writefln;

void main() {
    int[int] aa1 = [1:2, 3:4];
    int[int] aa2 = [5:6, 7:8];
    byte[int[int]] s;
    writefln(aa1, " ", aa2); // Prints: [1:2,3:4] [5:6,7:8]

    s[aa1] = 1;
    s[aa2] = 2;
    writefln(s); // Prints: [[1:2,3:4]:2]
}

Bye,
bearophile
Mar 16 2008
parent reply "Jarrett Billingsley" <kb3ctd2 yahoo.com> writes:
"bearophile" <bearophileHUGS lycos.com> wrote in message 
news:frjrkc$18k9$1 digitalmars.com...

 Can you explain this to me?
 Is this yet another corner case (or bug?)?

 import std.stdio: writefln;

 void main() {
    int[int] aa1 = [1:2, 3:4];
    int[int] aa2 = [5:6, 7:8];
    byte[int[int]] s;
    writefln(aa1, " ", aa2); // Prints: [1:2,3:4] [5:6,7:8]

    s[aa1] = 1;
    s[aa2] = 2;
    writefln(s); // Prints: [[1:2,3:4]:2]
 }

 Bye,
 bearophile


And to confirm it, s.length is 1, not 2.  There were originally some bugs in 
std.format's array and AA handling, but this is definitely not an output 
bug.

I'd bugzilla it if I were you.
Mar 16 2008
parent reply bearophile <bearophileHUGS lycos.com> writes:
Jarrett Billingsley:

 There were originally some bugs in std.format's array and AA handling,
 but this is definitely not an output bug. I'd bugzilla it if I were you.


I think this is the 13th bug I have found so far :-)
I think this may be caused by the fact that AAs don't have an == defined yet
(and no one seem to care). I can use my equalAA() function for most purposes,
but I presume hashing needs an == too, and I can't fix that with an external
function.

Bye,
bearophile
Mar 17 2008
next sibling parent "Jarrett Billingsley" <kb3ctd2 yahoo.com> writes:
"bearophile" <bearophileHUGS lycos.com> wrote in message 
news:frlmm1$2see$1 digitalmars.com...

 Jarrett Billingsley:

 There were originally some bugs in std.format's array and AA handling,
 but this is definitely not an output bug. I'd bugzilla it if I were you.


 I think this is the 13th bug I have found so far :-)


Welcome to the club.  No one ever said the reference implementation was 
bulletproof ;)
Mar 17 2008
prev sibling parent reply Frits van Bommel <fvbommel REMwOVExCAPSs.nl> writes:
bearophile wrote:

 I can use my equalAA() function for most purposes, but I presume hashing needs
an == too, and I can't fix that with an external function.


Yeah, TypeInfo.{getHash,equals,compare} aren't implemented for 
TypeInfo_AssociativeArray, TypeInfo_Function and TypeInfo_Delegate in 
either Phobos or Tango. And all of those are required for the relevant 
types to work correctly as AA keys, IIRC.

Of course, you could implement them and attach the patch to the relevant 
bug report... which didn't exist yet.
Remedied that last part:
Phobos: <http://d.puremagic.com/issues/show_bug.cgi?id=1926>
Tango: <http://www.dsource.org/projects/tango/ticket/988>
Mar 17 2008
parent bearophile <bearophileHUGS lycos.com> writes:
Frits van Bommel Wrote:

 Of course, you could implement them and attach the patch to the relevant 
 bug report... which didn't exist yet.
 Remedied that last part:
 Phobos: <http://d.puremagic.com/issues/show_bug.cgi?id=1926>
 Tango: <http://www.dsource.org/projects/tango/ticket/988>


You are good.
For reference this is the equalAA in my libs you can find online:


/***************************************
The == among two AAs.

DMD V.1.015 doesn't define < <= == != >= > among AAs.
(Python defines them all among dicts, so a full cmpAA can be defined too).
*/
bool equalAA(TyK1, TyV1, TyK2, TyV2)(TyV1[TyK1] aa1, TyV2[TyK2] aa2) {
    static if( !is(TyK1 == TyK2) || !is(TyV1 == TyV2) )
        return false;
    else {
        if (aa1.length != aa2.length)
            return false;

        foreach(k, v; aa1) {
            auto k_in_aa2 = k in aa2;
            if (!k_in_aa2 || (*k_in_aa2 != v))
                return false;
        }

        return true;
    }
}

unittest { // Test of equalAA()
    assert(  equalAA(AA!(int, int), AA!(int, int)) );
    assert( !equalAA(AA!(char, int), ['a':2, 'b':3]) );
    assert( !equalAA(['a':2, 'b':3], AA!(char, int)) );
    assert( !equalAA(['a':2, 'b':3], AA!(int, int)) );
    assert(  equalAA(['a':2, 'b':3], ['a':2, 'b':3]) );
    assert( !equalAA(['a':2, 'b':3], ['a':2]) );
    assert( !equalAA(['a':2, 'b':3], ['a':2, 'b':3, 'c':1]) );
    assert( !equalAA(['c':2, 'b':3], ['a':2, 'b':3]) );
    assert( !equalAA(['a':3, 'b':3], ['a':2, 'b':3]) );
    assert( !equalAA([2:'a', 3:'b'], ['a':2, 'b':3]) );
}

(AA!() creates an empty AA array).
Note it may not work if keys and/or values are static arrays (it can be
modified in an ugly way to work in such situation too)
(Note that is gives false if the items are of different type, even if they can
be cast-converted to each other. This may be different from arrays.)

Bye,
bearophile
Mar 17 2008