At 9:27 AM +0100 5/24/07, John Maddock wrote:
Kim Barrett wrote:
At 9:36 AM +0100 5/23/07, John Maddock wrote:
I've applied the patch to the 1.34 branch, applying the patch to cvs HEAD gives me some errors:
$ patch -p1 -i kab*.patch
I think your problem is right there. There were two patch files in the tarball:
kab-iostreams-RC_1_34_0.patch -- apply this to a sandbox that is on the RC_1_34_0 branch
kab-iostreams-trunk.patch -- apply this to a sandbox that is on HEAD.
Don't try to apply both patches to one place.
I wasn't, I extracted each file separately to different Boost trees. Only the trunk patch is present, but it still fails with the messages shown.
Hm. The trunk patch applies cleanly for me, both against the dated tree I created it against and against a fresh cvs tree from a few minutes ago. But I have a suspicion of what the problem might be. You performed tests against various windows compilers and cygwin-gcc. I'm going to guess that you are applying the patch on a windows machine, either under windows or under cygwin. I both developed the patch and just verified that it applied cleanly on linux. The two files that failed to patch cleanly for you both have RC_1_34_0 branch changes since HEAD that are being back ported to HEAD. Among the changes for both of these files are newline cleanups: http://boost.cvs.sourceforge.net/boost/boost/libs/iostreams/doc/tree/tree.cs... see version 1.4.4.1 http://boost.cvs.sourceforge.net/boost/boost/libs/iostreams/test/putback_tes... see version 1.4.4.1 I suspect some confusion in the patch creation / application process around the different newline representations is biting us here. I'm attaching a tarball containing the two files in question, post patch.