Closed renatosilva closed 9 years ago
We already dump the Windows version, and already dump DrMinGW versions in the report which are much less relevant. It makes no sense not to dump all module versions, when they might actually provide useful info (a wrong DLL version might cause a crash, and not appear in the stack trace very easily -- all it takes is returning a wrong result and cause the caller to crash -- and all you see in the stack trace the info about the caller).
Many other exception handlers dump all modules too.
I can consider swap the order (e.g, put modules version first, stack back trace last) if everybody thinks that would be better.
But omit restrict module versions seems going backwards to me.
Why don't make it a runtime option (on/off) like log file location?
I don't oppose the idea of adding runtime options to exchndl. But I still believe this is a good behavior.
The version-info dumping implemented in a36b428 floods the crash report with many unrelated DLLs. It was supposed to not clutter the stack trace, but it clutters the whole report instead. I suggest you to cache the involved modules in
dumpStack
for printing only these.