From mboxrd@z Thu Jan 1 00:00:00 1970 From: Benjamin Kosnik To: gcc@gcc.gnu.org Cc: law@redhat.com Subject: Re: GCC 3.2 Date: Mon, 12 Aug 2002 19:05:00 -0000 Message-id: <20020812190157.2508a7d6.bkoz@redhat.com> X-SW-Source: 2002-08/msg00716.html > You could certainly do this too. The idea behind having it done in > libstdc++ is that we can verify that sizes/layouts of any externally > visible classes and structures don't change from one rev of libstdc++ to > another. That in conjunction with testing the signatures of every > externally visible function gives us a much better chance of keeping > libstdc++ compatible from one rev to the next. Right. I think this is the preferred way to go, it just requires much (much much much) more work than simply swapping libstdc++.so's. It looks like LSB is also interested in some kind of solution like this, so perhaps there is some synergy there. -benjamin