www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.bugs - [Issue 5996] New: [CTFE] Undefined function call in auto return function

reply d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=5996

           Summary: [CTFE] Undefined function call in auto return function
           Product: D
           Version: D2
          Platform: x86
        OS/Version: Windows
            Status: NEW
          Keywords: diagnostic
          Severity: normal
          Priority: P2
         Component: DMD
        AssignedTo: nobody puremagic.com
        ReportedBy: bearophile_hugs eml.cc


--- Comment #0 from bearophile_hugs eml.cc 2011-05-13 13:10:26 PDT ---
This is wrong D2 code, because min() is not defined:


auto foo(int n) {
    uint two = 2, three = 3, five = 5;
    auto h = new typeof(two)[n];
    h[0] = 1;
    auto x2 = two, x3 = three, x5 = five;
    int i, j, k;
    for (int w = 1; w < n; w++) {
        auto el = h[w] = min(x2, x3, x5);
        if (x2 == el) x2 = two   * h[++i];
        if (x3 == el) x3 = three * h[++j];
        if (x5 == el) x5 = five  * h[++k];
    }
    return h[$ - 1];
}
enum uint h = foo(1691);
static assert(h == 2123366400);
void main() {}


DMD 2.053beta prints:
Assertion failure: '0' on line 1601 in file 'expression.c'

---------------------

If I have remove the three if() lines:


auto foo(int n) {
    uint two = 2, three = 3, five = 5;
    auto h = new typeof(two)[n];
    h[0] = 1;
    auto x2 = two, x3 = three, x5 = five;
    int i, j, k;
    for (int w = 1; w < n; w++) {
        auto el = h[w] = min(x2, x3, x5);
    }
    return h[$ - 1];
}
enum uint h = foo(1691);
static assert(h == 2123366400);
void main() {}


Now the error messages are:

Error: array index 4294967295 is out of bounds [][0 .. 0]
test2.d(10): Error: h[__dollar - 1u] cannot be interpreted at compile time
test2.d(12): Error: cannot evaluate foo(1691) at compile time
test2.d(12): Error: cannot evaluate foo(1691) at compile time
test2.d(13): Error: static assert  (foo(1691) == 2123366400u) is not
evaluatable at compile time

Note an error message without line number too.


DMD 2.052 instead prints:

test2.d(4): Error: array index 0 is out of bounds [][0..0]
test2.d(12): Error: cannot evaluate foo(1691) at compile time
test2.d(12): Error: cannot evaluate foo(1691) at compile time
test2.d(12): Error: cannot evaluate foo(1691) at compile time
test2.d(13): Error: static assert  (foo(1691) == 2123366400u) is not
evaluatable at compile time

---------------------

If I remove "auto" as return type and put a "uint":


uint foo(int n) {
    uint two = 2, three = 3, five = 5;
    auto h = new typeof(two)[n];
    h[0] = 1;
    auto x2 = two, x3 = three, x5 = five;
    int i, j, k;
    for (int w = 1; w < n; w++) {
        auto el = h[w] = min(x2, x3, x5);
    }
    return h[$ - 1];
}
enum uint h = foo(1691);
static assert(h == 2123366400);
void main() {}


Finally the error messages are correct:

test3.d(8): Error: undefined identifier min, did you mean function main?
test3.d(12): Error: cannot evaluate foo(1691) at compile time

--------------------------

Now adding the missing min() fixes the problem, and it works with auto return
type too:


T min(T)(T a, T b, T c) {
    return (a > b) ?   ((b > c) ? c : b)  :  ((a > c) ? c : a);
}
auto foo(int n) {
    uint two = 2, three = 3, five = 5;
    auto h = new typeof(two)[n];
    h[0] = 1;
    auto x2 = two, x3 = three, x5 = five;
    int i, j, k;
    for (int w = 1; w < n; w++) {
        auto el = h[w] = min(x2, x3, x5);
        if (x2 == el) x2 = two   * h[++i];
        if (x3 == el) x3 = three * h[++j];
        if (x5 == el) x5 = five  * h[++k];
    }
    return h[$ - 1];
}
enum uint h = foo(1691);
static assert(h == 2125764000);
void main() {}

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
May 13 2011
next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=5996


Don <clugdbug yahoo.com.au> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|diagnostic                  |ice-on-invalid-code
                 CC|                            |clugdbug yahoo.com.au


--- Comment #1 from Don <clugdbug yahoo.com.au> 2011-07-05 17:12:13 PDT ---
Reduced test case:
---------
auto bug5996() {
    if (anyOldGarbage) {}
    return 2;
}
enum uint h5996 = bug5996();
static assert(h5996 == 2);

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jul 05 2011
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=5996



--- Comment #2 from bearophile_hugs eml.cc 2011-07-06 03:20:22 PDT ---
(In reply to comment #1)

 Reduced test case:


My second example shows an error message with missing line number:
Error: array index 4294967295 is out of bounds [][0 .. 0]

I have reduced it to this, I think it's better to fix this before fixing your
reduced test case:

auto foo(int n) {
    auto h = new typeof(something)[n];
    return h[$];
}
enum uint f = foo(1);
void main() {}

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jul 06 2011
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=5996



--- Comment #3 from Don <clugdbug yahoo.com.au> 2011-07-06 04:21:21 PDT ---
(In reply to comment #2)

 (In reply to comment #1)

 Reduced test case:


 
 My second example shows an error message with missing line number:
 Error: array index 4294967295 is out of bounds [][0 .. 0]
 
 I have reduced it to this, I think it's better to fix this before fixing your
 reduced test case:
 
 auto foo(int n) {
     auto h = new typeof(something)[n];
     return h[$];
 }
 enum uint f = foo(1);
 void main() {}


It's a problem with the gagging system. It actually shouldn't get as far as
CTFE, so the out-of-bounds error shouldn't happen.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Jul 06 2011
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=5996



--- Comment #4 from github-bugzilla puremagic.com 2012-11-07 12:46:18 PST ---
Commit pushed to master at https://github.com/D-Programming-Language/dmd

https://github.com/D-Programming-Language/dmd/commit/baaa223916953667ef63c03448201bd1d18bcf48
Fix issue 5996 ICE(expression.c) CTFE of erroneous auto return function

When running semantic3 on a function to determine its return value, errors
must not be gagged. There is no chance to re-run semantic3 on it.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Nov 07 2012
prev sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=5996


Walter Bright <bugzilla digitalmars.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: -------
Nov 07 2012