On Tue, 25 Dec 2007, Felix Fietkau wrote: > Robert P. J. Day wrote: > > > it's obvious that, even if ccache is determined to be already > > installed, that Makefile is still going to cause the ccache source > > tarball to be downloaded, after which it's just going to be > > ignored. wouldn't it make more sense to check *immediately* if a > > particular (and, again, sufficiently recent) host tool is already > > on the system, at which point that entire Makefile could just be > > no-opped out of the way? that is, the basic call to "HostBuild" > > can just be avoided at that point, no? since you effectively have > > *nothing* to do.
> I agree. But simply adding some more code for preventing the > download is not the right solution. I'm thinking about writing some > generic code for handling that kind of packages, so the whole check > can be simplified a lot. I'll do it when I find the time for it. i probably don't know enough advanced make to be sure of this, but it would seem that this could be done with *less* code, not more. if you look at the number of build targets in tools/ccache/Makefile that need to be undefined when you realize you already have ccache: ... else define Build/Prepare endef define Build/Configure endef define Build/Compile endef define Build/Clean endef endif ... i would have thought a simpler solution would be to simply no-op the entire Makefile if you realize you already have an acceptable version of the tool installed. effectively, you'd end up just avoiding calling: $(eval $(call HostBuild)) just a thought, i'll look at this more closely myself. rday -- ======================================================================== Robert P. J. Day Linux Consulting, Training and Annoying Kernel Pedantry Waterloo, Ontario, CANADA http://crashcourse.ca ======================================================================== _______________________________________________ openwrt-devel mailing list [email protected] http://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
