[zeromq-dev] [PATCH] Throw noncontinuable exception on Win32 instead of abort

Paul Betts paul at paulbetts.org
Thu Oct 27 00:44:43 CEST 2011


> 1. The patch doesn't seem to print out the filename and the line number 
> of where the expection happened. This is crucial for identifying what 
> exactly happened.
I don't quite understand this - if you've got a debugger attached, the execution will stop at the assert instead of terminating, you'll have the full stack leading up to the crash, not just the file/line you asserted.  

> 2. Why use 0x40000015 instead of STATUS_FATAL_APP_EXIT?
STATUS_FATAL_APP_EXIT is defined in a later version of the Windows SDK (Vista I believe), and I didn't want to break XP. On XP, it will still work the same, but the number might not be identified by the debugger as such. 

-- 
Paul Betts <paul at paulbetts.org>


On Wednesday, October 26, 2011 at 4:33 AM, Martin Sustrik wrote:

> Hi Paul,
> 
> Two comments:
> 
> 1. The patch doesn't seem to print out the filename and the line number 
> of where the expection happened. This is crucial for identifying what 
> exactly happened.
> 
> 2. Why use 0x40000015 instead of STATUS_FATAL_APP_EXIT?
> 
> Martin





More information about the zeromq-dev mailing list