runit not buildable, only build-dependency is arch qualified
Debian autobuilders are currently refusing to build runit on many architectures (both official and unofficial) with the following message[1][2] runit build-depends on missing: - empty-dependency-after-parsing The immediate cause seems to be that runit recently removed procps from it's build-depends leaving the only build-dependency an arch qualified one, specifically. Build-Depends: dietlibc-dev (>> 0.28-0) [alpha amd64 arm hppa i386 ia64 mips mipsel powerpc ppc64 s390 sparc] Googling "empty-dependency-after-parsing" finds a single result which seems to be a patch to remove the code generating it from wanna-build along with some other code. [3] The patch description says that the removed functionality is now in dose but doesn't specifically say if the particular behviour on empty dependencies crossed over or not and a reply indicates that the patch may not actually have been commited. Comments in the code removed by the patch also hint that dpkg is also unhappy with this situation "At least as of now, empty is also an error from Dpkg::Deps, so better just prevent anything from building". So I decided to try a manual build on armhf (one of the affected architectures ). The package built successfully. Furthermore I don't see anything in policy that would forbid a package's only build dependency. Therefore IMO anything that fails to handle this case correctly is buggy. Do other people agree? if so where should wanna-build bugs be filed (since wanna-build is not currently in Debian)? does anyone have a simple testcase to see if the assertion in the comment about Dpkg::Deps is also true? [1] https://buildd.debian.org/status/package.php?p=runit&suite=sid [2] http://buildd.debian-ports.org/status/package.php?p=runit&suite=sid [3] https://lists.debian.org/debian-wb-team/2013/09/msg00033.html -- To UNSUBSCRIBE, email to debian-dpkg-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/53dacffa.40...@p10link.net
Re: runit not buildable, only build-dependency is arch qualified
On Fri, Aug 01, 2014 at 12:23:38AM +0100, peter green wrote: > Debian autobuilders are currently refusing to build runit on many > architectures (both official and unofficial) with the following > message[1][2] > runit build-depends on missing: > - empty-dependency-after-parsing > The immediate cause seems to be that runit recently removed procps from it's > build-depends leaving the only build-dependency an arch qualified one, > specifically. > Build-Depends: dietlibc-dev (>> 0.28-0) [alpha amd64 arm hppa i386 ia64 mips > mipsel powerpc ppc64 s390 sparc] > Googling "empty-dependency-after-parsing" finds a single result which seems > to be a patch to remove the code generating it from wanna-build along with > some other code. [3] The patch description says that the removed > functionality is now in dose but doesn't specifically say if the particular > behviour on empty dependencies crossed over or not and a reply indicates > that the patch may not actually have been commited. Comments in the code > removed by the patch also hint that dpkg is also unhappy with this situation > "At least as of now, empty is also an error from Dpkg::Deps, so better just > prevent anything from building". > So I decided to try a manual build on armhf (one of the affected > architectures ). The package built successfully. Furthermore I don't see > anything in policy that would forbid a package's only build dependency. > Therefore IMO anything that fails to handle this case correctly is buggy. Do > other people agree? Technically correct. OTOH, an empty build-dependency field means the package is not using debhelper for its build, but instead has some grotesque by-hand debian/rules file. So this error will only ever affect packages that are not using modern best-practice packaging. I'm sure I could easily find a volunteer to submit a patch switching runit to use dh(1), to make this issue go away for you. > if so where should wanna-build bugs be filed (since wanna-build is not > currently in Debian)? does anyone have a simple testcase to see if the > assertion in the comment about Dpkg::Deps is also true? There is a buildd.debian.org virtual package in the BTS: http://bugs.debian.org/buildd.debian.org Cheers, -- Steve Langasek Give me a lever long enough and a Free OS Debian Developer to set it on, and I can move the world. Ubuntu Developerhttp://www.debian.org/ slanga...@ubuntu.com vor...@debian.org signature.asc Description: Digital signature
Re: runit not buildable, only build-dependency is arch qualified
On Thu, Jul 31, 2014 at 04:46:32PM -0700, Steve Langasek wrote: > Technically correct. OTOH, an empty build-dependency field means the > package is not using debhelper for its build, but instead has some grotesque > by-hand debian/rules file. So this error will only ever affect packages > that are not using modern best-practice packaging. I'm sure I could easily > find a volunteer to submit a patch switching runit to use dh(1), to make > this issue go away for you. But I don't want that! I love "grotesque", ancient things that work more than 10 years with nearly no maintenance effort ;). Regards, Gerrit. -- To UNSUBSCRIBE, email to debian-dpkg-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/20140801052048.15031.qm...@40b5b75f24d749.315fe32.mid.smarden.org
