Norm Jacobs wrote: > George Vasick wrote: >> Norm Jacobs wrote: >>> Raj Prakash wrote: >>>> >>>> This information is Copyright 2009 Sun Microsystems >>>> 1. Introduction >>>> 1.1. Project/Component Working Name: >>>> GCC4: The GNU Compiler Collection 4.X >>>> >>>> 4. Technical Description: >>>> 4.1. Details: >>>> Commands will be installed in /usr/bin with versioned suffixes, >>>> e.g. gcc-4.3.2. The runtime libraries will be installed >>>> /usr/lib with major, minor, and patch suffixes as appropriate >>>> along with a link for the major version, e.g >>>> libstdc++.so.6.0.10 and libstdc++.so.6 -> libstdc++.so.6.0.10. >>>> See section 4.5 Interfaces below for additional details. >>>> >>>> This case proposes to modify the previous release, >>>> LSARC/2008/776 GNU Developer Collection, as follows: >>>> >>>> 1) Localized message files will be moved from /usr/share/locale >>>> to /usr/lib/gcc/<machine>/<version>/share/locale. >>>> >>>> 2) Runtime libraries will be refactored from a single package >>>> into multiple packages, one package per library, to allow >>>> individual libraries to be upgraded in future releases. >>> Did I understand correctly that you are refactoring the packaging so >>> that you can potentially deliver different pieces of gcc from >>> different releases of gcc in the future? >> >> This is actually to allow for the future possibility of the major >> version of one of the libraries changing. For example, GCC 4 provides >> libone.so.6 and libtwo.so.4 while GCC 5 provides on libone.so.7 and >> libtwo.so.4. If we made a single package containing all of the >> runtimes for each release as we do now, there would be a duplicate >> pathname, libtwo.so.4, between gccruntime4 and gccruntime5. >> Separating the runtime libraries into individual packages based on >> their major version numbers would avoid this problem. >> > So what happens when GCC5 libtwo.so.4 differ GCC4 libtwo.so.4 due to bug > fix or worse? It may be that they never differ or never make > incompatible change without bumping the version. I just want to know > that it's not going to be a problem.
The latter. This is where we count on committed stability for the libraries, i.e. no incompatible changes without bumping the major version. George > > -Norm > >
