Hi *, I spent 4 hours I did not have on this gs problem now and I don't have any ideas left. The problem is that currently the gs package is built in the build target of debian/rules but it is rebuilt from scratch in the install target for reasons I don't grok. The install target is run like this to get gs installed into a temporary dir: $(MAKE) prefix=`pwd`/debian/$(PACKAGE)/usr install I know I can probably work around this problem but I am quite sure that other problems will appear if I just override GS_LIB in the Makefile. So if anybody could spent some time and thoughts on this it would me most appreciated. My last idea with touching ld.tr did not work either - here is my previous mail to the bts: ----- Forwarded message from Torsten Landschoff <torsten@debian.org> ----- From: Torsten Landschoff <torsten@debian.org> To: Rene Engelhard <rene@debian.org> Cc: 173493@bugs.debian.org, ruckerz@verizon.ne Subject: Re: Bug#173493: latest unstable gs package has wrong location for gs_init.ps Hi Rene, On Wed, Dec 18, 2002 at 09:23:01AM +0100, Rene Engelhard wrote: > > I wonder where it got that path from. Something is definitely broken :( > > Funny thing is that I changed nearly nothing in the package. > > Anyway, as there _is_ a bug finding that it's not that important ;) I agree there is a bug. But currently I can't fix it - I don't grok what's going wrong there. The problem boils down to make rebuilding gs when running make install. As the gs Makefile does not support DESTDIR, I use $(MAKE) prefix=`pwd`/debian/$(PACKAGE)/usr install to install the build result. The joke is that this command rebuilds all of gs. I don't know why, it didn't with 7.05-1. And I neither changed the Makefile nor debian/rules - all files that differ from -1 to -2 are debian/control, debian/changelog and src/gdevpsim.c. I tried downgrading make but that did not make a difference. Now I am trying my luck with make -d, but I still can't understand what's going on. This pass I touch obj/ld.tr as that file triggers the whole rebuild because a prerequisite seems to be newer (not sure if make thinks the same instant is newer). If you ask me after a successful build make should make sure that the result has a newer time than the prerequisites (e.g. wait a second if the build was too fast and touch the result). I'm curious about the result of this build :) Greetings Torsten ----- End forwarded message -----
Attachment:
pgpkYtzJHtt64.pgp
Description: PGP signature