[lsb-discuss] deprecating some interfaces for LSB 3.2?

[Nick Stoughton]

On Mon, 2007-10-22 at 17:45 -0700, Wichmann, Mats D wrote:
> Currently when compiling LSB checkers, three
> interfaces cause gcc depraction warnings to
> be issued.
> 
> Should those three interfaces also move to
> deprecated status in LSB?
> 
> pthread_h.c: In function 'pthread_h':
> pthread_h.c:838: warning: 'pthread_attr_setstackaddr' is deprecated
> (declared at /usr/include/pthread.h:346)
> pthread_h.c:840: warning: 'pthread_attr_getstackaddr' is deprecated
> (declared at /usr/include/pthread.h:338)
> 
> unistd_h.c: In function 'unistd_h':
> unistd_h.c:2803: warning: 'getwd' is deprecated (declared at
> /usr/include/unistd.h:483)
> 
> 
> Note this isn't from a terribly recent gcc, either -
> 4.1.0 as found in SLES10.
> 
These interfaces are utterly non-portable, were marked as obsolete in
SUSv3, and have been removed from SUSv4.

However, as an ABI standard, we *could* specify these in the various
arch-dependent books, with a lot of work. Nevertheless, they are pretty
rarely used in real code, and have much better replacements
(pthread_attr_[sg]etstack). I would lean strongly in favor of
deprecating them. It is not an option to simoly say they are "as
described in SUSv3", since SUSv3 doesn't even make them implementation
defined, but leaves all the questions (like which direction does the
stack grow in) unspecified.
 
> This isn't a question of trying to toss these interfaces
> any time soon, but rather to match what the implementations
> we're trying to describe do... and be in a position to drop
> them if the implementations do decide to finally toss them
> (not expected any time soon).
> 
> 
> 
> _______________________________________________
> lsb-discuss mailing list
> lsb-discuss at lists.linux-foundation.org
> https://lists.linux-foundation.org/mailman/listinfo/lsb-discuss