On Sat, 4 Sep 2010 at 14:22:58 +0200, Andreas Metzler wrote: > On 2010-09-01 "Carlos R. Mafra" <[email protected]> wrote: > > On Wed, 1 Sep 2010 at 19:24:41 +0200, Andreas Metzler wrote: > >> Carlos R. Mafra <[email protected]> wrote: > >>> On Sun, 29 Aug 2010 at 13:07:50 -0700, Doug Barton wrote: > > >>>> As for symbol versioning, it can be useful, but is probably overkill > >>>> for something like libwraster unless there are plans to change its > >>>> internals dramatically down the road. I do agree that not exporting > >>>> private symbols is a good idea. > > >>> Is there an easy way to get the benefits of patch 3 (ie not export > >>> private symbols) without the versioning stuff? > > >> I think I could strip down the .map file to only list the the public > >> symbols and use > >> libtool -export-symbols =$(srcdir)/libwraster.sym > >> instead of > >> libtool -Wl,--version-script=$(srcdir)/libwraster.map > > >> I can come up with a patch (probably not before saturday) if that is > >> wanted. > > > Yep, and then the first two patches can be dropped from 'next'. > > Patch attached. This goes on top of
Thanks, will push it soon. > > git revert 7e5b8e412f8cdbf2a5865ba38ede1cc233cf731c > git revert 44edeea804f61e0e13bfa4f90f751395ac04d263 > > (I would prefer if these two commits were reverted in the repo instead > of pruned from the history.) May I ask why? That defeats the purpose of having the 'next' branch, which is exactly to raise discussions while the patches are there. When some agreement is met, patches will be dropped/redone etc. I was actually satisfied that this was the first time I would do this :-) And I am inclined to keep my satisfaction, unless you have a strong reason to keep two commits followed by two reverts. But as you are the author of said patches, I wouldn't like to overcome your wishes. -- To unsubscribe, send mail to [email protected].
