Re: [pdf-devel] Patch for FS#93 stack based iterators

[Michael Gold]

On Tue, Jan 27, 2009 at 19:14:35 -0800, address@hidden wrote:
>  > > OTOH, I'm really against assertions in code. I consider it a serious 
> desi=
>  > gn
>  > > flaw.
>  > 
>  > Why?
> 
> They cause the program to abort instead of giving the chance to handle the
> error.

There's no sensible way to handle something like a badly-defined struct,
though. In this case a unit test works just as well, since it doesn't
depend on any runtime state, but it seems reasonable to use asserts for
other problems that can only be caused by programmer error -- they
produce messages that are useful to programmers (rather than crashes or
undefined behaviour), and are easily disabled by defining NDEBUG.

> So, let's sum up. We use a 'void*' and add a test case with a:
> assert(sizeof(gl_itr)<=sizeof(our_itr))

Yes, but if it's in a unit test, fail_unless can be used instead of
assert.

> Any other suggestion ?

I can't think of anything else now.

-- Michael

signature.asc
Description: Digital signature