On Thursday, January 20, 2011 09:36:13 am Eric Noulard wrote: > 2011/1/20 David Cole <[email protected]>: > > Moving to the CMake developer's list, as requested by this bug comment: > > http://public.kitware.com/Bug/view.php?id=11693#c24958 > > > > Comments or thoughts on the topic are welcome... > > > > Please reply here with any further discussion before adding more info > > to the bug report. Once a consensus is reached here, we'll update the > > bug again. > > I'm not a Mac user and I'm a poor Windows user :-] > > Now on linux (or other unices) I was puzzled by the /usr/share/cmake-2.8 > thing in fact the vast majority unix software don't come with such version > mangling. FHS doesn't seems to include recommandation in this area: > http://www.pathname.com/fhs/. > > Sometimes the version mangling appears when a major update (python2 --> > python3) arrives and one needs to have both versions installed for > compatibility reason. In this case, most of the time, the up-to-date > version is unnumbered and the old compatibility one is renamed with > version mangling. > In fact before the big old->new update thing the new version may be > mangled for a > while waiting for wider acceptance (Python case). > > Some libs do always have name mangling because the version is really part > of their name., e.g. glib-2.0. > > Then update-alternative (Debian, Fedora, ?others?) may be used > to create appropriate links: > http://www.debian-administration.org/articles/91 > > So I vote for no version mangling. > Or may be an option (default to OFF==no mangling) in order to ease the > mangling of cmake-2.8 > when cmake 3.0 will be out :-]
I also vote for no version number, but allow the user to change it if they want. It seems that would give the general user a more clear upgrade path. Clint _______________________________________________ cmake-developers mailing list [email protected] http://public.kitware.com/cgi-bin/mailman/listinfo/cmake-developers
