I finally started to mess around with D and managed to crash the latest dmd with this test program:

struct bittest_T(int C)
{
 bit b[C];
};
void main()
{
 for (int t=0; t<32; t++)
  printf("sizeof bittest_T(%i) %i\n", t, bittest_T!(t).sizeof );
}

DMD prints this compile error (which makes sense):

C:\UTIL\d\fart.d(33): template instance fart.bittest_T!(t) cannot use local
't'
as template parameter

..and crashes with an access violation exception.
I ran DMD.EXE from MSDEV and it breaks at 0040B6D1:

0040B6CA 90                   nop
0040B6CB 90                   nop
(function entry point?)
0040B6CC 53                   push        ebx
0040B6CD 31 C0                xor         eax,eax
0040B6CF 89 CB                mov         ebx,ecx
0040B6D1 88 00                mov         byte ptr [eax],al

Pretty obvious: EAX is set to 0, and then used in a look-up? Memory ECX points to:

003C4438  C4 46 4A 00 B4 CD 39 00 03 00 00 00 1B 00 00 00  ÄFJ.´Í9......... 003C4448  60 40 39 00 1C 00 00 00 58 3B 39 00 00 00 00 00  `@9.....X;9..... 003C4458  7C 44 3C 00 00 00 00 00 00 00 00 00 00 00 00 00  |D<.............

Hope this helps,

Lionello.

-- Get the CACert root certificate (and a personal one) at http://cacert.org/

Same crash with this program:

void main()
{
 for (int t=0; t<33; t++)
  printf("sizeof bittest_T(0) %i\n", (bit[t]).sizeof );
}

(Guess what I'm trying to do :-S )

Memory ECX points to:

003C3EE0  C4 46 4A 00 B4 CD 39 00 04 00 00 00 1B 00 00 00  ÄFJ.´Í9......... 003C3EF0  60 40 39 00 1C 00 00 00 D0 3A 39 00 00 00 00 00  `@9.....?:9..... 003C3F00  24 3F 3C 00 00 00 00 00 00 00 00 00 00 00 00 00  $?<.............

L.

On Tue, 30 Nov 2004 09:23:26 +0200, Lionello Lunesu <lionello.lunesu@crystalinter.remove.com> wrote:

> Same crash with this program:
>
> void main()
> {
>  for (int t=0; t<33; t++)
>   printf("sizeof bittest_T(0) %i\n", (bit[t]).sizeof );
> }
>
<snip>

D should complain with:
Integer constant expression expected instead of t
and does for me.

-- 
"Unhappy Microsoft customers have a funny way of becoming Linux,
Salesforce.com and Oracle customers." - www.microsoft-watch.com:
"The Year in Review: Microsoft Opens Up"

"Clearly, if I were one of that Predator hunting team, I would file a
complaint with my local member asking why such an unsafe
environment should be allowed to continue, tradition or not."
- http://www.sydneyanglicans.net/culture/watching/alien_vs_predator/

I just tested again, it really crashes.

# void main()
# {
#  for (int t=0; t<33; t++)
#   printf("sizeof bittest_T(0) %i\n", (bit[t]).sizeof );
# }

DMD 0.107 doesn't say a thing; crashes immediately. Doesn't print the link-line either so I guess it's the compiler.

Running WindowsXP with SP2:

Microsoft Windows XP [Version 5.1.2600]
Digital Mars D Compiler v0.107
OPTLINK (R) for Win32  Release 7.50B1

L.

(I've even tried running dmd from cmd.exe, 4nt.exe 5.00u and msdev6: no difference)

(XP created this file. Is this useful?)

[D:\] dmd -v crash.d
parse     crash
semantic  crash
semantic2 crash
semantic3 crash

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


Added to DStress as http://svn.kuehne.cn/dstress/nocompile/array_initialization_04.d

Thomas

PS: I can't reproduce this under Linux.

Lionello Lunesu schrieb am Tue, 30 Nov 2004 16:20:44 +0200:
> I just tested again, it really crashes.
>
> # void main()
> # {
> #  for (int t=0; t<33; t++)
> #   printf("sizeof bittest_T(0) %i\n", (bit[t]).sizeof );
> # }
>
> DMD 0.107 doesn't say a thing; crashes immediately. Doesn't print the link-line either so I guess it's the compiler.
>
> Running WindowsXP with SP2:
>
> Microsoft Windows XP [Version 5.1.2600]
> Digital Mars D Compiler v0.107
> OPTLINK (R) for Win32  Release 7.50B1
>
> L.
>
> (I've even tried running dmd from cmd.exe, 4nt.exe 5.00u and msdev6: no difference)



-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.9.9 (GNU/Linux)

iD8DBQFBrXwg3w+/yD4P9tIRAqtsAJ9J0MDaCdhErUc73xdURyfBUZCqxgCfZFtb
nY6rdW358w97XSQUTL+jz+4=
=VqLd
-----END PGP SIGNATURE-----