Re: when can a package be made architecture-dependent?
On Fri, 17 Jan 2003, Manoj Srivastava wrote:
> On Fri, 17 Jan 2003, Bill Allombert wrote:
> >> Debian developer reference 5.3.1 states that
> >> 2. Don't set architecture to a value other than ``all'' or ``any''
> >> unless you really mean it. In too many cases, maintainers don't
> >> follow the instructions in the Debian Policy Manual
> >> (http://www.debian.org/doc/debian-policy/). Setting your
> >> architecture to ``i386'' is usually incorrect.
>
> Steven> That's good to know, except what are the instructions of the
> Steven> Debian Policy Manual on the subject? I couldn't find any.
>
> I suppose they instructions are "listen to the developers
> reference"? Policy is trying to diet down to a minimalist set of
> hard rules; the developers reference is the compendium of best
> practices.
Okay, slimming down policy makes sense. In that case, though, the
developers' reference should probably eliminate this pointer to the Policy
Manual since it leads in a circle, and maybe be more explicit with what
"really mean it" means. How about:
2. Don't set architecture to a value other than ``all'' or ``any''
unless the upstream package is intrinsically unportable
(e.g. a program to disable a Pentium CPU ID). If the package
is theoretically portable, even if it currently fails to build on
some architectures, it should be set to architecture any/all to
open a path for future porters. Setting your architecture to
``i386'' is usually incorrect.
Steven
Reply to: