Bug#396435: subversion: FTBFS: test switch_tests.py fails

[Peter Samuelson]

severity 396435 minor
thanks

[Lucas Nussbaum]
> Not allowing to build as root is a bad idea.

Why do you, or anyone, need to build packages as root?  It's a bad idea
in general.  For example, a former version of Samba actually gets
incorrect 'configure' values if configure is run as root.

> You should detect that the build is running as root and ignore the
> result of this specific test, not just fail like that.

I'm adding this to debian/rules:

        @if [ $$(id -u) = 0 ] && [ -z "$(FAKEROOTKEY)" ]; then \
                echo >&2 "***"; \
                echo >&2 "*** Building as root is not supported ***"; \
                false; \
        fi

This will: (1) tell the user exactly what is wrong so she knows how to
fix it; (2) tell the user that the issue is known and need not be
reported; (3) by failing early, avoid wasting an hour of build time.

Note that I'm special-casing fakeroot.  Believe it or not, fakeroot
doesn't fail here because it doesn't emulate access(2).

You still shouldn't need to build packages under fakeroot, though.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://lists.alioth.debian.org/pipermail/pkg-subversion-maintainers/attachments/20061031/1a59d2d5/attachment.pgp