On Thu, Jun 14, 2007, Jeff Johnson wrote:
> On Jun 14, 2007, at 8:11 AM, Ralf S. Engelschall wrote:
>
>> I have one subtle issue on my RPM/POPT todo list since a longer
>> time: POPT's use of the ancient 32V AT&T UNIX alloca(3), a machine-,
>> compiler-, and system-dependent and especially non-POSIX (and hence
>> less portable) function whose use is even discouraged on lots of Unix
>> platforms.
>>
>
> There are those who like alloca and those who don't. ;-)
>
> Please cite your personal preferences/concerns rather than the arch
> "ancient" and "discouraged".
My personal preference is to avoid any non-POSIX functions in small
stand-alone libraries like POPT. In RPM (where we need huge Autoconf
glues and have already many system specifics) alloca(3) is fine. I just
prefer to keep those reusable libraries like POPT extremely clean by
using really just standardized API functions whenever possible.
> [...]
> As long as "make check" passes, all changes to popt are fair game.
> [...]
Puhhh... I usually at least would avoid like hell to change the external
and well-established API of POPT ;-) That's why I just concentrate on
build environment, portability and internal cleanups here.
>> Any objections?
>
> Nope.
Thanks.
Ralf S. Engelschall
[EMAIL PROTECTED]
www.engelschall.com
______________________________________________________________________
POPT Library http://rpm5.org
Developer Communication List [email protected]
