Re: RFC 140 (v1) One Should Not Get Away With Ignoring System Call Errors

Jarkko Hietaniemi wrote:
> 
> I don't think we should assume that we can't scratch our nose without
> getting overly object-happy.  For the particular aspect I'm driving at
> there is no *need* for objects of any kind, no catch, no throw, no
> structured exceptions: I want the program simply to _die_.  I'm not
> against someone else providing higher level ways handle/trap/propagate
> those croakings, be the way procedural or OO or what is the rave du
> jour.  Don't get me wrong: for some purposes and somtimes I like
> structured warnings.  But what I am proposing is very low-level, it's
> below the level of Perl the language, actually.  Think of it as a
> cleaned-up consistent wrapping of the current C library APIs.  If
> somebody wants to further wrap these failures into something nicer,
> okay, I have no problem with that.

Agreed.  I can even imagine that system functions would, via die,
put some simple data structure in $@ and @@, but whenever these
values are read, the are seamlessly and internally promoted to
first-class Exception objects.  This way, the guts see internal
exceptions as simple data structures, and the user code sees
all exceptions as simple objects.

Yours, &c, Tony Olekshy