[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#178809: rules for Build-Depends-Indep satisfaction make no sense



On Tue, Jan 28, 2003 at 08:11:34PM +0100, Bas Zoetekouw wrote:
> Package: debian-policy
> Version: 3.5.8.0
> Severity: important
> 
> Currently, policy says that following about Build-Depends-Indep (section
> 7.6): 
> 
> | The Build-Depends-Indep and Build-Conflicts-Indep fields must be
> | satisfied when any of the following targets is invoked: build, clean,
> | build-indep, binary and binary-indep.
> 
> This makes no sense, because it would mean that Build-Depends-Indep
> dependencies would have to be installed anyway (for clean target for
> example) even when building only the arch-dependent binary packages.

Sorry, it is correct.  See the footnote to that section.  If you're
building the arch-dependent binary packages using binary-arch, then
you don't need the Build-Depends-Indep packages, and if you do, then
you have a FTBFS serious bug.

> | <ifvoid> The Build-Depends-Indep and Build-Conflicts-Indep fields must
> |    be satisfied when any of the following targets is invoked: build, 
> |    clean, build-indep, binary and binary-indep.
> | <ifvoid> note that that includes build

Correct.

> | <elmo> policy is broken
> | <BenC> way broken
> | <ifvoid> ok
> | <BenC> that's just stupid
> | <BenC> that pretty much says that indep needs to always be satisfied, 
> |    which makes indep useless

No: if binary-arch depends (in a Makefile sense) on build, then you're
not actually "invoking" build, and your make can do what it likes, as
long as you only need the Build-Depends packages.  If you make build,
then you should require both Build-Depends and Build-Depends-Indep.  I
know that's not what the autobuilders yet do, but one day they might
check for the existence of the build-arch target, and fall back to a
build target if that doesn't exist.  At that point, the distinction
will make sense; the way the Build-Depends{,-Indep} fields were
originally designed or implemented was fundamentally broken, in that
the -Indep fields were useless.

   Julian

-- 
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

        Julian Gilbey, website: http://www.polya.uklinux.net/
   Debian GNU/Linux Developer, see: http://people.debian.org/~jdg/
     Visit http://www.thehungersite.com/ to help feed the hungry



Reply to: