On Thu, Apr 16, 2015 at 11:34 PM, Peter Maydell <[email protected]> wrote:
> > You don't need to, because the makefile sets PYTHON. > (It's possible Makefile guesses wrongly; you could argue for > making it make a more sophisticated guess, like "python2 unless > there's no such executable in which case try plain python".) > In fact, I tried the Makefile way after reading t/README : The easiest way to run tests is to say "make -C t" but that didn't work : make: Entering directory 'stgit/t' test.py make: test.py: Command not found Makefile:10: recipe for target 'all' failed make: *** [all] Error 127 make: Leaving directory 'stgit/t' That's where I'm coming from... The explanation why that didn't work is that make -C t is doing the same as "cd t && make" manually (i.e. outside from any other Makefile) which does not get the exported PYTHON from the top-level Makefile, as it would if run as a recursive make invokation. > > I debated with myself adding: > > > > PYTHON ?= python2 > > > > to the makefile. > > That line is already in Makefile (and since commit e22723c > we correctly pass it down and honour it in all the sub-makefiles). > Yes, it does when running "make test" from top level directory, but not in case you do it manually. I'd rather it work in all cases. That's why I came with that patch. So we alternatively could fix the t/README... What do you think ? -- Vincent Legoll
_______________________________________________ stgit-users mailing list [email protected] https://mail.gna.org/listinfo/stgit-users
