Re: glibc recompiling was Re: libc resolver problem solved (critical bug)

[Oleg Krivosheev <kriol@fnal.gov>]

   On Thu, 19 Nov 1998, Dale Scheetz wrote:

   > My inclination is to "move along" to the new libraries ASAP, rebuilding
   > those packages that were built with -5. Any other step looks like lost
   > ground that will eventually have to be covered, so why not do it now?

   As I noted in one of my previous posts, please add a conflict against
   current libstdc++ because this new libc will break all c++ apps linked
   against libstdc++2.9 (i'm not sure if it is just the library or the apps
   too :( ), like apt, dselect, etc...
   And please add conflicts to broken dpkg and apt too so we dont get a
   broken package management system. I guess then these packages will have to
   be compiled with dpkg from hamm.

   And please remove the version information from the .shlibs file, it will
   no longer be needed, and this will help to identify the packages compiled
   against the wrong libc... all of those will probably have to be recompiled
   :( who is going to submit the bugreports ? 

   So the idea should be: you compile this libc, someone compiles egcs with
   stuff from hamm, so we get a working libstdc++, then dpkg and apt.
   Needs coordinated work...

Exactly my feeling...

my believe is in order to get out of this we have to upload
statically linked apt/dpkg, then fix glib/egcc/libstdc++ and
then come back with dynamic apt/dpkg back.

   Btw, CC goes to Brian, because this is a very serious release issue...

   -- 
   Madarasz Gergely           gorgo@caesar.elte.hu         gorgo@linux.rulez.org

OK