dpkg --print-architecture [WAS: Release-critical Bugreport for January 25, 1999]

["Marcelo E. Magallon" <mmagallo@efis.ucr.ac.cr>]

On Mon, Jan 25, 1999 at 12:15:11AM -0600, BugScan reporter wrote:

> Package: emacs20 (main)
> Maintainer: Rob Browning <rlb@cs.utexas.edu>
>   28177  dpkg --print-architecture requires gcc                                 
> 
> Package: xlib6 (main)
> Maintainer: Branden Robinson <branden@debian.org>
>   31610  xlib6: requires gcc but declares no dependency (dpkg --print-gnu-build-architecture?)

Aren't this two the same kind of bug?  Having no idea why the maintainer
chose this particular way to implement this, wouldn't it be easier to put
this information *on the script* at *compile-time*?

Something like

ARCH := $(shell dpkg --print-gnu-build-architecture)

debian/whatever.postinst: debian/whatever.postinst.in
	sed -e 's/@ARCH@/$(ARCH)/' $< > $@

Are there any drawbacks with this approach?  If I'm reading the scripts
correctly, the intent is to find out the architecture of the host that will
be *using* the packages, not the architecture of the host where the packages
are being *installed* (it would be interesting to know how/if cross-arch
installations are feasible, they _are_ possible via --force-architecture)


						Marcelo