On Thu, 2017-09-14 at 13:40 +0100, Daniel P. Berrange wrote: > On Thu, Sep 14, 2017 at 12:58:26PM +0100, Peter Maydell wrote: > > On 14 September 2017 at 12:55, Gerd Hoffmann <[email protected]> > > wrote: > > > Hi, > > > > > > > I think a better approach is to have something in rules.mak > > > > that ensures the submodule is checked out correctly (only > > > > when building from GIT, not dist), and then have the rules > > > > which generate the keymap files depend on this. > > > > > > Care sending a patch doing that for dtc? > > > > It sounds awfully fiddly. Maybe it is the best we can do > > given the mess that is git submodules, but is it really > > the common approach? > > I'll do a prototype so we can see something concrete working and > evaluate how pleasant (or not) it is
Tried to brew something: https://www.kraxel.org/cgit/qemu/log/?h=work/submodule dtc was pretty simple due to the recursive make call. Hooking the submodule update into a non-recursive make looks complicated, especially because the submodule update might change the timestamps and therefore the target set which needs a rebuild ... So I did the keymaps build with a recursive make call too, which doesn't look that pretty ... cheers, Gerd
