Re: [AD] About create_bitmap

[Peter Wang <tjaden@xxxxxxxxxx>]

Chris wrote:

> On Wed, 03 Nov 2004 18:31:02 +0100, Elias Pschernig <elias@xxxxxxxxxx> wrote:
>  
>
> > Well, you could use a simple #define. The vast majority of uses of
> > create_bitmap will not need a check - so this is a very special use.
> > There's of course no problem with create_bitmap, nobody would notice it.
> > But in general, I don't think it is a good idea. Else, we end up having
> > error checking for every single function, like OpenGL.
> >    
>
> I don't see error checking as a bad thing. It's not hard to do, and in
> fact is already done via ASSERT. What's so bad about changing an
> ASSERT (which applies only to debugmode) to if (which is always
> checked)? The function is not speed-critical, so the more error
> checking the better.
I'm afraid I'm to blame for this.  I didn't like (and still don't like) 
functions being lenient about incorrect arguments -- incorrect arguments 
should cause a function to blow up, not be silently corrected.  Anyway, 
I might have taken it too far in this case.  Checking for negative 
arguments to create_bitmap() is probably ok, since it would return NULL 
anyway and the program unlikely to continue very far.

(Apologies to Lennart, who asked me about this a couple of years ago.)

> And I think it would be more useful than you
> think, since there are several situations where you just obtain
> numbers and then pass them directly to create_bitmap.
>  
This is still a bad idea if you're not sure about the numbers you 
obtained.  You might have code that "works" in release mode, but not in 
debug mode (as it shouldn't).  Or do you mean to get rid of the ASSERTs 
completely and replace them with if's?

Peter