On Fri, 30 Apr 2010 21:00:24 +0200, Tomas Carnecky <tom@dbservice.com> wrote: > Like on Mac OS X, the linker doesn't automatically resolve > dependencies. Thanks for this fix, Tomas! I like the fix described above, but you seem to have sent a patch with three independent changes, only one of which was described. See below for what I've done... > notmuch: $(notmuch_client_modules) lib/libnotmuch.a > - $(call quiet,CXX $(CFLAGS)) $^ $(FINAL_LIBNOTMUCH_LDFLAGS) -o $@ > + $(call quiet,CXX $(CFLAGS)) $^ $(FINAL_NOTMUCH_LDFLAGS) -o $@ That seems to be correct. Could you describe what problem you ran into that motivated you to find this fix? I have not committed this piece. > -printf "Checking for Mac OS X (for shared library)... " > +printf "Checking which platform we are on... " > if [ `uname` = "Darwin" ] ; then > - printf "Yes.\n" > - mac_os_x=1 > + printf "Mac OS X.\n" > + platform=MACOSX ... > -ifeq ($(MAC_OS_X),1) > +ifeq ($(PLATFORM),MACOSX) > LIBRARY_SUFFIX = dylib > # On OS X, library version numbers go before suffix. > LINKER_NAME = libnotmuch.$(LIBRARY_SUFFIX) That hunk was all described by your original commit message. So I've applied this part with your original commit message and will be pushing it soon. > +/* Helper functions to test if a given dirent is of a certain type > + */ > +static int > +_is_reg(const char *path, struct dirent *entry) ... > @@ -372,7 +426,7 @@ add_files_recursive (notmuch_database_t *notmuch, > > if (! S_ISREG (st.st_mode)) > continue; > - } else if (entry->d_type != DT_REG) { > + } else if (!_is_reg(path, entry)) { > continue; > } This all might be well and good, but is also not described by the commit message. So I have not committed this either. If you resend with an actual description of the change, I'll be happy to take a closer look. Thanks again, -Carl -- carl.d.worth@intel.com