PATCH: fix sleb128 for 64 vs 32 host

[Tristan Gingold]

> On 19 Nov 2015, at 06:08, Alan Modra <amodra@gmail.com> wrote:
> 
> On Mon, Nov 16, 2015 at 12:44:48PM +0100, Tristan Gingold wrote:
>> Is it really worth stripping leading digits in output_big_?leb128 ?
> 
> Yes, I think we must do that..
> 
>> The only operation performed on bignum is the negation.
> 
> ..due to this operation (and bitwise not) leaving leading sign bits.

Ok, but:

1) Do you agree the initial case is a bug ?

2) If so, do you think it is worth fixing it ?

3) If so, how ?  
   Should we keep a sign bit and not extending the bignum for ‘-‘ and ‘~’ ?

Tristan.