Re: Special symbol version for glibc's internal interfaces

["Karl J. Runge" <karl dot runge at sun dot com>]

Hi Andreas,

Thanks for the feedback on the special symbol version for
private/internal interfaces.

I'm guessing Ulrich sent a pointer to you WRT our "abicheck" project
that discusses this idea and provides an example tool that checks an
application's conformance (I also really like the rpm --requires
mechanism :-)

In case not, here is the pointer to the preview area of the project
while we are getting feedback from library developers like you:

	http://abicheck.sourceforge.net/preview/

The username is "abicheck" and the passwd is "abc".

Thanks,

Karl

On Tue, 28 Aug 2001, Andreas Jaeger <aj@suse.de> wrote:
> 
> Karl Runge and David Brown propose in [1] following their experience
> with Solaris, to add a special symbol version for glibc's internal
> implementation interface, e.g. GLIBC_PRIVATE.
> 
> For example, debug/Versions contains currently:
>     # functions used in other libraries
>     __backtrace; __backtrace_symbols; __backtrace_symbols_fd;
> 
> Hiding those symbols with GLIBC_PRIVATE would make it explicit that
> those are internal interfaces.
> 
> Why didn't we follow this at first when adding symbol versioning to
> glibc?  What do other think about this?
> 
> Andreas

-----------------------------------------------------------------------------
Karl J. Runge                                     Sun Microsystems, Inc.     
Staff Engineer, ABI Program                       P.O. Box 4001
e-mail:   karl.runge@sun.com                      Burlington, MA  01803-0902
Phone:    (781) 442-2366                          Building 2, One Network Dr.
FAX:      (781) 442-1437                          Mailstop UBUR02-201