| Re: [AD] 4.3 error handling |
[ Thread Index |
Date Index
| More lists.liballeg.org/allegro-developers Archives
]
> > On Thursday 08 June 2006 2:34 pm, Peter Hull wrote:
> > > I agree with Bob;l I don't think we should do this. I can't actually
> > > see the problem that these proposed features are intended to solve.
It solves the problems of newbies not checking return values (so instead of
erroring several lines down unexpectedly, it throws up a helpful error
immediately; something even veteran programmers can benefit from when they
just want to code something fast), and it solves the problem of having to
check each and every function that may or may not fail, even functions
you're "sure" can't fail, and functions you don't want to check for speed
reasons.
On Thursday 08 June 2006 13:56, Elias Pschernig wrote:
> Yes, the only thing that was doubted is the setjmp/longjmp - the TLS
> error code and enums and everything else of course is very nice.
And ironicly, it may be the enums and TLS that cause the biggest problem. ::)
Currently, some of the enum names clash with OpenAL, and the TLS may or may
not throw a big "surprise!" when the lib is loaded dynamicly (via LoadLibrary
or dlopen). That last point I'm not so sure about with C99's __thread, the
specs don't seem to hint at any problems regarding this, but Windows' TLS
seems to be more complicated (as usual).