http://d.puremagic.com/issues/show_bug.cgi?id=1253

           Summary: DMD 0.175 introduced bug: array initializers as
                    expressions are not allowed?
           Product: D
           Version: 0.175
          Platform: All
        OS/Version: All
            Status: NEW
          Keywords: rejects-valid
          Severity: major
          Priority: P2
         Component: DMD
        AssignedTo: bugzilla@digitalmars.com
        ReportedBy: oskar.linde@gmail.com


void main() {
    struct T { int a; int b;}
    const T[] arr = [{1,1},{2,2}];
    T[] arr2 = arr.dup;
    assert(arr2 == arr);
}

Gives:

test.d:3: Error: array initializers as expressions are not allowed test.d:3: Error: array initializers as expressions are not allowed

Interestingly, commenting out the assert line makes the code compile. changing const -> static also makes the code compile.

The bug is confirmed to exist on 0.175, 0.176, 0.178, 1.006, 1.007, 1.013 (gdc
0.23) and 1.014

On DMD 0.174 (and earlier), the code compiles ok.


-- 

http://d.puremagic.com/issues/show_bug.cgi?id=1253


smjg@iname.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |smjg@iname.com
           Severity|major                       |regression




-- 

http://d.puremagic.com/issues/show_bug.cgi?id=1253





------- Comment #1 from thomas-dloop@kuehne.cn  2007-08-25 16:42 -------
Added to DStress as http://dstress.kuehne.cn/run/s/struct_initialization_11_A.d http://dstress.kuehne.cn/run/s/struct_initialization_11_B.d http://dstress.kuehne.cn/run/s/struct_initialization_11_C.d


-- 

http://d.puremagic.com/issues/show_bug.cgi?id=1253





------- Comment #2 from terranium@yandex.ru  2008-05-11 10:38 -------
exists in 2.012, doesn't depend on asserts

struct A
{
  int a,b;
}

void f()
{
  A[2] rg=[{1,2},{1,2}];
}


-- 

http://d.puremagic.com/issues/show_bug.cgi?id=1253





------- Comment #3 from terranium@yandex.ru  2008-05-11 10:51 -------
workaround :-/

struct A
{
        int a,b;
}

void f()
{
        A[2] rg=[A(1,2),A(3,4)];
}


-- 

http://d.puremagic.com/issues/show_bug.cgi?id=1253





------- Comment #4 from terranium@yandex.ru  2008-09-30 04:47 -------
other initialization-related info
bug 2096, bug 2170, bug 2356, bug 2375


-- 

http://d.puremagic.com/issues/show_bug.cgi?id=1253


Don <clugdbug@yahoo.com.au> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |clugdbug@yahoo.com.au
            Version|0.175                       |2.022
            Summary|DMD 0.175 introduced bug:   |array initializers as
                   |array initializers as       |expressions are not allowed
                   |expressions are not         |in const arrays
                   |allowed?                    |
           Severity|regression                  |normal




--- Comment #5 from Don <clugdbug@yahoo.com.au>  2009-09-01 07:33:56 PDT ---
This works for me on D1.046. It was fixed somewhere between 1.020 and 1.041.

It also works on DMD2.031, if you change 'const' to 'enum' to preserve the semantics.

Oddly, however, on DMD2.031 the code as-written still displays:

bug.d(173): Error: array initializers as expressions are not allowed

So this is now a D2-only bug, which is not a regression. It's never worked with these semantics. Checked as far back as 2.022.

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

http://d.puremagic.com/issues/show_bug.cgi?id=1253


Don <clugdbug@yahoo.com.au> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |patch




--- Comment #6 from Don <clugdbug@yahoo.com.au>  2009-09-01 08:24:13 PDT ---
And here's a patch.
Cause: The initializer is used in declaration.c, VarDeclaration::semantic().
We see the lines:
------
            Expression *e = init->toExpression();
            if (!e)
            {
            init = init->semantic(sc, type);
            e = init->toExpression();
            if (!e)
------
The initializer is supposed to return NULL if the initializer semantic hasn't been run yet. But the array initializer doesn't -- it prints an error, and returns an ErrorExp instead! It never has a chance.

PATCH: init.c, last lines of ArrayInitializer::toExpression() (around line
473).

Lno:
    delete elements;
-   error(loc, "array initializers as expressions are not allowed");
-   return new ErrorExp();
+    return NULL;
}

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

http://d.puremagic.com/issues/show_bug.cgi?id=1253


Walter Bright <bugzilla@digitalmars.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
                 CC|                            |bugzilla@digitalmars.com
         Resolution|                            |FIXED


--- Comment #7 from Walter Bright <bugzilla@digitalmars.com> 2009-10-06 02:19:51 PDT ---
Fixed dmd 2.033

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