Re: [boost] low-level sub-allocation/alignment challenge

From: David Abrahams (abrahams_at_[hidden])
Date: 2000-08-24 13:13:10

----- Original Message -----
From: "Greg Colvin" <greg_at_[hidden]>

> So this is a POD too?
>
>   struct Rule
>   {
>      unsigned long cost;
>      const void* action;
>      char& operator[](size_t i) { return ((char*)(&action+1))[i]; }
>      char operator[](size_t i) const { return ((char*)(&action+1))[i]; }
>  };

Yes. But your array is an array of char; mine was an array of char*, FWIW.

> > > And if action were of type char instead of
> > > const void*, then your symbols() array would most likely be
misaligned.
> >
> > Correct.  In that case you'd need to do something extra to ensure
alignment.
>
> Why?  Doesn't char have minimal alignment?

Exactly. And the expression &action + 1 would be aligned on a 1-byte
boundary not, e.g. a 4-byte boundary as probably needed to access it as an
array of pointers.

-Dave