[pacman-dev] [PATCHv2 2/2] handle arch specific attributes

[Dave Reisner]

On Sun, Aug 03, 2014 at 06:50:19PM +1000, Allan McRae wrote:
> On 03/08/14 01:54, Dave Reisner wrote:
> > This introduces support for architecutre-specific conflicts, depends,
> > optdepends, makedepends, replaces, and conflicts by appending "_$CARCH"
> > to the array name. For example, in the global section:
> > 
> >   arch=('i686' 'x86_64')
> >   depends=('foo')
> >   depends_x86_64=('bar')
> > 
> > This will generate depends of 'foo' and 'bar' on x86_64, but only 'foo'
> > on i686. Moreover, this is supported in the package functions with the
> > same heuristics as the generic names, e.g.
> > 
> >   ...
> >   arch=('i686' 'x86_64')
> >   depends=('foo')
> >   ...
> > 
> >   package_somepkg() {
> >     depends_x86_64=('bar')
> > 
> >     ...
> >   }
> > 
> > Again, will cause x86_64 to have depends of 'foo' and 'bar', but only
> > 'foo' for i686.
> > ---
> > v2:
> > 1) extended this to a bunch more attributes -- some of them I can't
> > necessarily reason an immediate use case for, but conceptually could
> > have their uses as arch-specific attributes.
> > 2) Added documentation
> > 3) Added check_sanity support post mega-refactoring.
> > 
> >  doc/PKGBUILD.5.txt    | 21 +++++++++++++++++++++
> >  scripts/makepkg.sh.in | 46 +++++++++++++++++++++++++++++++++++++++++++---
> >  2 files changed, 64 insertions(+), 3 deletions(-)
> > 
> > diff --git a/doc/PKGBUILD.5.txt b/doc/PKGBUILD.5.txt
> > index 17e8af2..5cebbbd 100644
> > --- a/doc/PKGBUILD.5.txt
> > +++ b/doc/PKGBUILD.5.txt
> > @@ -183,17 +183,26 @@ If the dependency name appears to be a library (ends with .so), makepkg will
> >  try to find a binary that depends on the library in the built package and
> >  append the version needed by the binary. Appending the version yourself
> >  disables automatic detection.
> > ++
> > +Architecture-specific depends can be added by appending an underscore and the
> > +architecture name e.g., 'depends_x86_64=()'.
> >  
> 
> Do we want to repeat this sentence many times or add a single section
> detailing architecture specific additions.

I think it's more clear to document it directly alongside the relevant
variables instead of breaking out a new section.

> Also, despite "added" there, I do not find it clear that these are
> overrides.  Perhaps "Additional architecture-specific..." would help.
> If this was in its own section, it could be expanded for clarity.
> 

It's not clear that they're overrides because they *aren't* overrides.
If you specify:

  arch=('i686' 'x86_64')
  depends=('libfoo')
  depends_x86_64=('lib32-libbar')

Then, you create an i686 package with depends of 'libfoo' only, and an
x86_64 package with depends of 'libfoo' *and* 'lib32-libbar'.