Hi John, * John Calcote wrote on Tue, Mar 04, 2008 at 10:38:50PM CET: > Newbie question here: Is it better to do this in a Makefile.am: > > install-exec-hook: > $(mkdir_p) $(DESTDIR)@syslogdir@ > > or this: > > install-exec-hook: > $(mkdir_p) $(DESTDIR)$(syslogdir) > > assuming, of course, that I called AC_SUBST(syslogdir) in configure.ac? > > The first will expand to $(prefix)/..., so you'd have the option of > changing the resulting directory by passing prefix= on the make > command line. The second will allow you to change the resulting > directory with either prefix= or syslogdir= on the make command line. > I guess you'd get a bit more flexibility by using the second option > because you could change the syslogdir without changing the prefix. > But, is there a conventional, or more correct approach?
Nice question. You already deliver the answer, including full rationale and all: the second form is preferable because it allows more override. In general, there is one other bit of difference between @foo@ and $(foo) in that automake may not be able to look into the former. But for this case that makes little difference. Cheers, Ralf
