Hi, (dropping lists out of To: as discussion died there and doesn't need project wide audience IMO)
Sorry, missed your message in the thread: On Sat, Dec 08, 2007, Stefano Zacchiroli wrote: > > PYVERS := $(shell pyversions -vr debian/control 2>/dev/null) > > > > build-%: > > PYTHON=`which python$*` ./configure > > I guess this should have been "`which $*` ./configure", right? If so, > read on. Nope, this is correct, "pyversions -vr debian/control" gives you "2.4 2.5", not "python2.4 python2.5". (In practice, I don't think there are many cases where one must be used instead of the other; it just happened that in some cases I or others preferred building for all python versions instead of for all python interpreters; at random or as a matter of taste.) > > And it doesn't even conflict with implemeting build-arch; it just makes > > the test make moot. > And what would you loose in using something like: > build-python%: > PYTHON=`which python$*` ./configure > ... > build-arch: > ... I wouldn't lose anything; I don't think I would gain anything either; the point was that it isn't a good idea to call make as a method to check whether I implemented the build-arch concept. In fact, I already went on and changed my packages to use build-stamp/% or similar instead (build-%/build-stamp, build-stamp-%, build-%/configure-stamp etc., depending on the package.) and hence avoid this bug. Bye, -- Loïc Minier
