From: Peter Dimov (pdimov_at_[hidden])
Date: 2003-01-13 15:18:20
From: "William E. Kempf" <wekempf_at_[hidden]>
>
> If the exception type doesn't fold multiple errors into a single unit,
> there's no need for the error code in this situation. RTTI will provide
> the same capabilities, even if you don't want to have seperate catch
> clauses.
It won't, unless you are willing to create a dedicated exception class for
each and every potential OS error code. :-)
The idea is that at the throw point, you are converting an OS error code to
a portable error identification token (exception type, portable error code,
string identifier, whatever) and this conversion is, in general, lossy. So
it might make sense to preserve the original.