• BCS (15/15) Sep 23 2006 This *is* a bug. Has it been reported yet?
• BCS (2/2) Sep 23 2006 forgot to mention
• Mikola Lysenko (6/26) Sep 24 2006 Probably a new discovery. This is related to the try-catch bug that has...
• BCS (4/10) Sep 24 2006 Likely this is related, but both of them only mention cases where delega...
• Mikola Lysenko (12/16) Sep 25 2006 From what I've worked out, it appears to be an issue with the way the

BCS <nowher spamdrop.uidaho.edu> writes:

This *is* a bug. Has it been reported yet?

<code>
import std.stdio;

void main()
{
	{
		scope(exit) writef("a\n");
	}

	scope(exit) writef("b\n");
}
<code>

output:

a
b
b

BCS <nowhere pamdump.uidaho.edu> writes:

forgot to mention

DMD 0.164 linux

Mikola Lysenko <mclysenk mtu.edu> writes:

BCS wrote:
 This *is* a bug. Has it been reported yet?
 
 <code>
 import std.stdio;
 
 void main()
 {
 	{
 		scope(exit) writef("a\n");
 	}
 
 	scope(exit) writef("b\n");
 }
 <code>
 
 output:
 
 a
 b
 b


Probably a new discovery.  This is related to the try-catch bug that has 
been around since DMD 0.150.  So far, no fix.  Here are the relevant posts:

http://lists.puremagic.com/pipermail/digitalmars-d-bugs/2006-May/000854.html
http://lists.puremagic.com/pipermail/digitalmars-d-bugs/2006-August/001911.html

Hopefully they will get fixed soon.

BCS <nowhere spamdump.uidaho.edu> writes:

== Quote from Mikola Lysenko (mclysenk mtu.edu)'s article
 Probably a new discovery.  This is related to the try-catch bug that
 has been around since DMD 0.150.  So far, no fix.  Here are the
 relevant posts:

http://lists.puremagic.com/pipermail/digitalmars-d-bugs/2006-May/000854.html

http://lists.puremagic.com/pipermail/digitalmars-d-bugs/2006-August/001911.html
 Hopefully they will get fixed soon.

Likely this is related, but both of them only mention cases where delegates are
used, How does mine create a delegate? (From the scope statement?) OTOH is this
bug known to cause problems elsewhere? (I didn't read then whole thread)

Mikola Lysenko <mclysenk mtu.edu> writes:

BCS wrote:
 Likely this is related, but both of them only mention cases where delegates are
 used, How does mine create a delegate? (From the scope statement?) OTOH is this
 bug known to cause problems elsewhere? (I didn't read then whole thread)
 

 From what I've worked out, it appears to be an issue with the way the 
compiler calculates offsets for things like catch and finally blocks in 
general.  The delegates are just one case of this bug.  Whenever there 
is more than one set of scope exit blocks, the bug will manifest.

It appears to be because the compiler does not properly compute the 
offsets in one of the DHandler structs (though the exact details escape 
me at the moment, since it has been a few months since I spent any time 
wrestling with this problem.)  It is a rather aggravating problem, which 
took me several hours of single stepping through the asm from deh2.d to 
figure out what was going.

I certainly hope it gets fixed in the next release.