The "official" Redhat/Fedora rpms for boost-1.32 had set soname as e.g.: libboost_date_time.so.1 instead of libboost_date_time.so.1.32 I wonder if I should do the same for 1.33? Does anyone know if the API is compatible from 1.32 to 1.33? Or, has Redhat made a mistake in doing this?
Neal Becker wrote:
The "official" Redhat/Fedora rpms for boost-1.32 had set soname as e.g.: libboost_date_time.so.1 instead of libboost_date_time.so.1.32
I wonder if I should do the same for 1.33? Does anyone know if the API is compatible from 1.32 to 1.33?
I'm sure it's not. Even such small library as program_options has binary incompatible changes from 1.32 to 1.33 (for example, removed method). I suspect that larger libraries has lots of such changes.
Or, has Redhat made a mistake in doing this?
Maybe. - Volodya
Vladimir Prus wrote:
Neal Becker wrote:
The "official" Redhat/Fedora rpms for boost-1.32 had set soname as e.g.: libboost_date_time.so.1 instead of libboost_date_time.so.1.32
I guess that answers the question I had in the other thread. That's a mistake sure to lead to chaos. The default system layout build of BBv1 sets the soname to libboost_date_time.so.1.33.0 and my hope was that this would be used by the RPMs. But go figure that they would override that, it's the GNU way to impose their view on others :-( Yes, at minimum the soname should have "1.33" in it and the RPMs should be fixed. My preference is to have the full "1.33.0" in it, but I'm paranoid when it comes to version compatibility. -- -- Grafik - Don't Assume Anything -- Redshift Software, Inc. - http://redshift-software.com -- rrivera/acm.org - grafik/redshift-software.com -- 102708583/icq - grafikrobot/aim - Grafik/jabber.org
participants (3)
-
Neal Becker -
Rene Rivera -
Vladimir Prus