Re: [bug #39334] obstack code is non portable

[David Monniaux]

On 06/24/2013 10:46 PM, Paul Eggert wrote:
> Perhaps in theory, but if it works on all practical environments then
> it's good enough for now. If there's a practical environment where it
> doesn't work, please let us know so that we can fix it for that platform. 

What happened is that we were testing the front-end of a static analysis
system and ended up with some pointer arithmetic on NULL in the middle
of a function. We found that very weird, especially since this portion
of the code was apparently reachable.

You are correct that this should not break on most current environments
(AS/400 is dealt with separately), but we thought that this was worth
filing because it's the kind of issue that can arise mysteriously when
compiling with a new compiler or a new version thereof, or moving to a
new platform (recall all the problems with 64-bit cleanliness).

-- 
David Monniaux, directeur de recherche au CNRS
Laboratoire VERIMAG http://www-verimag.imag.fr/~monniaux/