[Build-common-hackers] Bug#489430: Bug#489430: closed by Jonas Smedegaard <dr at jones.dk> (Bug#489430: fixed in cdbs 0.4.53)

[Martin Koeppe]

On Wed, 18 Mar 2009, Frédéric Brière wrote:

> On Wed, Mar 18, 2009 at 09:48:32PM +0100, Martin Koeppe wrote:
>> The behaviour still hasn't changed. time_1.7-23 can't still be built
>
> Hmm, me thinks Jonas mistook r269 for something else.  :)
>
>> with debhelper's dh_testroot, instead the hardcoded
>>   test "`id -u`" = 0
>> is executed.
>
> What intrigues me is that dh_testroot does (almost) just that:
>
>  if ($< != 0) {
> 	error("You must run this as root (or use fakeroot).");
>  }
>
> Why would one work but not the other?

That's not the point here. rules includes debhelper.mk, so the 
maintainer apparently intends to use debhelper. If he had included 
debhelper.mk first, this would be the case. Unfortunately, many 
maintainers don't include debhelper.mk first as they should.

This bug is about 'wrong' inclusion order in rules, and 
the current 0.4.53 still can't handle the 'wrong' order, so the bug 
shouldn't be closed.

But FYI: On interix root does not have uid 0 (no POSIX requirement), 
and I have patched debhelper for that. So for me it would be important 
that dh_testroot is used (I think one could expect that from 
debhelper.mk anyway), so I don't need to patch cdbs also and still 
can use it with the many packages that use debhelper.mk, even if they 
use the wrong inclusion order. For more details see also #416018 and 
http://www.debian-interix.net

The post from Jonas on 2008-07-12 is about using functions, not 
conditionals, but apparently this has not been done.

> Notice that dh_testroot tests for the *real* UID, while buildcore.mk
> tests for the *effective* UID.  Does this explain the discrepancy?
> (What does "id -ru" return for you?)

No, it doesn't matter, see above.


Martin