Re: stack module

[Marc Nieper-Wißkirchen]

Hi Bruno,

> In Gnulib, we usually avoid extensive use of multi-line macros, because
> it hampers debuggability. Here, however, one needs macros, in order to
> accommodate the TYPE argument, which is not necessarily compatible to 'void 
> *'.
> Nevertheless, we would try to put as much code as possible into functions.
> The STACK_INIT macro, for example, could be implemented as a function.

How? This would mean that the function stack_init has to take a void *
argument, which has to be cast to

struct
{
  void *base; ...
}

and we have to hope that this structure is guaranteed to be compatible to

struct
{
  type *base; ...
}

by the C standard.

> > #define STACK_CLEAR(stack)            \
> >   free ((stack).base)
>
> Shouldn't this one also set .size and .allocated to 0 ?

A stack can be uninitialized or initialized. An uninitialized stack is
initialized by STACK_INIT. It is cleared (and uninitialized) by
STACK_CLEAR. An uninitialized stack does not have to maintain any
invariant. The only way to use it is to initialize it again. Thus,
setting .size and .allocated seems pointless.

> > #define STACK_POP(stack)            \
> >   (stack).base [--(stack).size]
> >
> > #define STACK_DISCARD(stack)            \
> >   (--(stack).size)
> >
> > #define STACK_TOP(stack)            \
> >   (stack).base[(stack).size - 1]
>
> In these three macros, I would consider to abort() when (stack).size == 0.

In the form of assure of the assure module? Or, to facilitate
optimization, better assume from verify module? In non-debug builds, I
want to make sure that no superfluous checks are made.

Marc