Re: Exceptions vs. undefined behaviour

Johan Torp ha scritto:
> IIRC, the standard library avoids throwing exceptions so that users
> which compile without exception sup****t shall be able to use as much
> of the standard library as possible. Is this correct, incorrect or are
> there other motivations?

That is not entirely correct. There are several situations where the
library is expected/required to throw an exception. In fact there is a
whole hierarchy of exception cl***** for that purpose. The library
avoids throwing exceptions in all places where "failure" is not
considered an "error" (for example: std::find) and in few other places.
There are also a few places where exceptions are not thrown and
behaviour is left undefined in case of error. This is usually done for
sake of performances (for example: std::vector<>::operator[]) or to
avoid over-constraining the implementation (std::tr1::shared_from_this).

> Will this also be the case for the newly accepted libraries and for
> TR2?

It's likely that the same approach described above will be followed for
TR2.

> Will this "undefined behaviour"-strategy only apply to broken pre-
> conditions or will error codes and the like be used for re****ting
> internal errors?

There is no "undefined behaviour"-strategy. The standard clearly states
when the behaviour is defined and when it isn't. If the behaviour is
defined the standard clearly states if an exception shall be thrown or
if an error condition is re****ted in some other way. If the behaviour is
undefined, which is usually the case when a precondition is not
satisfied, anything can happen: the implementation might ignore or
re****t the problem, an exception might be thrown or the application
might crash or anything else.

HTH,

Ganesh

-- 
      [ See http://www.gotw.ca/resources/clcm.htm
for info about ]
      [ comp.lang.c++.moderated.    First time posters: Do this! ]