digitalmars.D.bugs - [Issue 8903] New: Bad code for enum array members
- d-bugmail puremagic.com (96/96) Oct 28 2012 http://d.puremagic.com/issues/show_bug.cgi?id=8903
- d-bugmail puremagic.com (13/13) Oct 28 2012 http://d.puremagic.com/issues/show_bug.cgi?id=8903
- d-bugmail puremagic.com (14/14) Oct 28 2012 http://d.puremagic.com/issues/show_bug.cgi?id=8903
- d-bugmail puremagic.com (10/10) Jan 31 2013 http://d.puremagic.com/issues/show_bug.cgi?id=8903
http://d.puremagic.com/issues/show_bug.cgi?id=8903
Summary: Bad code for enum array members
Product: D
Version: D2
Platform: All
OS/Version: All
Status: NEW
Keywords: wrong-code
Severity: normal
Priority: P2
Component: DMD
AssignedTo: nobody puremagic.com
ReportedBy: siegelords_abode yahoo.com
struct A
{
immutable int arr[8] = [1, 2, 3, 4, 5, 6, 7, 8];
}
struct B
{
enum int arr[8] = [1, 2, 3, 4, 5, 6, 7, 8];
}
void test_a(int n)
{
auto a = A();
auto b = a.arr[n];
}
/*
0000000000416c68 <_D4test6test_aFiZv>:
416c68: 55 push rbp
416c69: 48 8b ec mov rbp,rsp
416c6c: 48 83 ec 20 sub rsp,0x20
416c70: 89 7d f8 mov DWORD PTR [rbp-0x8],edi
416c73: c6 45 f0 00 mov BYTE PTR [rbp-0x10],0x0
416c77: 48 63 45 f8 movsxd rax,DWORD PTR [rbp-0x8]
416c7b: 48 83 f8 08 cmp rax,0x8
416c7f: 48 89 45 e8 mov QWORD PTR [rbp-0x18],rax
416c83: 72 0a jb 416c8f <_D4test6test_aFiZv+0x27>
416c85: bf 10 00 00 00 mov edi,0x10
416c8a: e8 a1 00 00 00 call 416d30 <_D4test7__arrayZ>
416c8f: 48 8b 4d e8 mov rcx,QWORD PTR [rbp-0x18]
416c93: 8b 14 8d f0 41 63 00 mov edx,DWORD PTR [rcx*4+0x6341f0]
416c9a: c9 leave
416c9b: c3 ret
*/
void test_b(int n)
{
auto a = B();
auto b = a.arr[n];
}
/*
0000000000416c9c <_D4test6test_bFiZv>:
416c9c: 55 push rbp
416c9d: 48 8b ec mov rbp,rsp
416ca0: 48 83 ec 20 sub rsp,0x20
416ca4: 89 7d f8 mov DWORD PTR [rbp-0x8],edi
416ca7: c6 45 f0 00 mov BYTE PTR [rbp-0x10],0x0
416cab: 48 63 45 f8 movsxd rax,DWORD PTR [rbp-0x8]
416caf: 48 b9 08 00 00 00 00 movabs rcx,0x8
416cb6: 00 00 00
416cb9: 48 3b c1 cmp rax,rcx
416cbc: 48 89 45 e8 mov QWORD PTR [rbp-0x18],rax
416cc0: 72 0a jb 416ccc <_D4test6test_bFiZv+0x30>
416cc2: bf 16 00 00 00 mov edi,0x16
416cc7: e8 64 00 00 00 call 416d30 <_D4test7__arrayZ>
416ccc: 48 be 08 00 00 00 00 movabs rsi,0x8
416cd3: 00 00 00
416cd6: 48 bf 60 43 63 00 00 movabs rdi,0x634360
416cdd: 00 00 00
416ce0: e8 1b 23 00 00 call 419000 <_d_arrayliteralTX>
416ce5: c7 00 01 00 00 00 mov DWORD PTR [rax],0x1
416ceb: ba 02 00 00 00 mov edx,0x2
416cf0: 89 50 04 mov DWORD PTR [rax+0x4],edx
416cf3: c7 40 08 03 00 00 00 mov DWORD PTR [rax+0x8],0x3
416cfa: c7 40 0c 04 00 00 00 mov DWORD PTR [rax+0xc],0x4
416d01: c7 40 10 05 00 00 00 mov DWORD PTR [rax+0x10],0x5
416d08: c7 40 14 06 00 00 00 mov DWORD PTR [rax+0x14],0x6
416d0f: c7 40 18 07 00 00 00 mov DWORD PTR [rax+0x18],0x7
416d16: c7 40 1c 08 00 00 00 mov DWORD PTR [rax+0x1c],0x8
416d1d: 48 8b 4d e8 mov rcx,QWORD PTR [rbp-0x18]
416d21: 8b 04 88 mov eax,DWORD PTR [rax+rcx*4]
416d24: c9 leave
416d25: c3 ret
416d26: 90 nop
416d27: 90 nop
*/
void main()
{
}
It makes no sense to me for the enum array to cause re-allocation when
instantiated. I know that enum arrays are problematic entities to begin with,
but while they are allowed to exist, they shouldn't generate horrible code like
that.
--
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Oct 28 2012
d-bugmail puremagic.com
http://d.puremagic.com/issues/show_bug.cgi?id=8903
yebblies <yebblies gmail.com> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |RESOLVED
CC| |yebblies gmail.com
Resolution| |DUPLICATE
This is not wrong-code, this is a performance issue.
*** This issue has been marked as a duplicate of issue 2356 ***
--
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Oct 28 2012
http://d.puremagic.com/issues/show_bug.cgi?id=8903
siegelords_abode yahoo.com changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|RESOLVED |REOPENED
Resolution|DUPLICATE |
No, it is not. Issue 2356 does not affect the array in struct A. The bug here
is the difference between A and B and not how badly B is implemented. Both
cases (or should) involve static array literal initialization, with the bug
being that B does not use it.
--
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Oct 28 2012
http://d.puremagic.com/issues/show_bug.cgi?id=8903 I don't think the code should compile. 'enum' should be declaring a manifest constant, it shouldn't be possible to index it at run time. The whole point of these enum constants was that they shouldn't appear in the executable! If you want it to appear in the executable, use can use immutable or const. This is a duplicate of another bug, I forget which one though. -- Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email ------- You are receiving this mail because: -------
Jan 31 2013