Bug#817033: sumo: FTBFS when built with dpkg-buildpackage -A (in a random way)

[Santiago Vila]

Hi. Found this:

Makefile.am says:

pydoc:
        rm -rf docs/pydoc
        mkdir docs/pydoc
        cd docs/pydoc && \
        for i in `find ../../tools/traci ../../tools/sumolib -name "*.py" -not -executable | sed 's,../../tools/,,;s,/,.,g;s,.py,,;s,.__init__,,'`; do \
            PYTHONPATH="../../tools" python -c "import $$i, pydoc; pydoc.writedoc($$i)"; \
        done

so the order in which the different "python -c" commands are executed
depends on the order in which "find" outputs its result, but this can be
anything because there is no canonical filesystem ordering.

In the end, the build fails or it does not depending on the *last*
executed command.

For example, this simple Makefile will fail:

all:
	true; false
	echo Hi

and this one will succeed:

all:
	false; true
	echo Hi


If we are going to forgive failure in any of the "python -c" commands,
maybe adding "|| true" at the end will do.

OTOH, if every "python -c" command should work flawlessly, the
current Makefile does not correctly express such condition.

Thanks.