• "Matthew Wilson" <dmd synesis.com.au> Dec 23 2002
• "Matthew Wilson" <dmd synesis.com.au> Dec 23 2002
• "Matthew Wilson" <dmd synesis.com.au> Dec 23 2002

Walter

Am getting some builds that cause

  Internal error: cgobj 386
  --- errorlevel 1

on the 8.32b, and not on 8.31. Also have some that cause the ICE on both
8.31/8.32b, but not on earlier, say 8.28.

Will try and isolate (although the workload is making this look pretty
unlikely)

Presume you can't act until I narrow it, yes?

More research has shown that this is a combinatorial bug.

The certain piece of code that, when commented out, stops the ICE only
causes it in one project (where it is used in combination with a number of
other headers/templates), and does not cause the same problem in its own
(smaller) test program.

I am, therefore, out of ideas, beyond sending you the code for both projects
(2 .cpp, 11 .h). Shall I?

Matthew


"Matthew Wilson" <dmd synesis.com.au> wrote in message
news:au83nm$on5$1 digitaldaemon.com...
 Walter

 Am getting some builds that cause

   Internal error: cgobj 386
   --- errorlevel 1

 on the 8.32b, and not on 8.31. Also have some that cause the ICE on both
 8.31/8.32b, but not on earlier, say 8.28.

 Will try and isolate (although the workload is making this look pretty
 unlikely)

 Presume you can't act until I narrow it, yes?

I've done a workaround, and made it go away. I've attached the header in
case you're interested:  look for _STLSOFT_COMPILER_IS_DMC.

I would think it would be something you would like to track down, since it
is a vulnerability that has crept in sometime between 8.28 and 8.31, but I
guess it can go down on the not-urgent list.

Matthew


"Matthew Wilson" <dmd synesis.com.au> wrote in message
news:au83nm$on5$1 digitaldaemon.com...
 Walter

 Am getting some builds that cause

   Internal error: cgobj 386
   --- errorlevel 1

 on the 8.32b, and not on 8.31. Also have some that cause the ICE on both
 8.31/8.32b, but not on earlier, say 8.28.

 Will try and isolate (although the workload is making this look pretty
 unlikely)

 Presume you can't act until I narrow it, yes?