[RFC] alternative bss approach

[Denis Vlasenko]

On Thursday 14 June 2007 17:20, Luciano Rocha wrote:
> On Thu, Jun 14, 2007 at 05:04:31PM +0200, Bernhard Fischer wrote:
> > Hi,
> > 
> > http://www.simtreas.ru/~dzo/common_bss.tgz
> > 
> > vodz said: This is a common minimal bss method for busybox without
> > malloc. I'm interrested in feedback about this approach.
> > 
> > </quote>
> > I didn't have a chance to look at it myself yet.
> > vda, perhaps it's interresting?
> > 
> 
> Eh? Why does it use magic comments and a script to calculate the size of

Even worse: what if I have a pointer member? What the size is?

> the structure (the common bss), when an union would give it for free?

union is nice when you don't have applet private types, and then
objects of these types (or pointers or arrays of those) in bss.

Take a look at ash to apprecite worst case scenario - when you have
applet so big and complex that it has subsections with their own
types, global variables etc...
--
vda