Hi Please ignore the patch debian-stubdom.Makefile.diff . Not needed if make in debian/rules is simply run as "env -u LDFLAGS make install-stubdom". Sorry for noise. Only 2 issues on xen-unstable.hg ... - xen-watchdog-bug.diff - obvious bug on all platforms - debian-python-2.6.diff - only needed for simple build on debian with no .deb packaging ... and one issue for debian packager: - "env -u LDFLAGS" must be used in debian/rules makefile on two commands: env -u LDFLAGS make install-xen env -u LDFLAGS make install-stubdom When the above is resolved xen-unstable.hg will be "fully debian-compatible". :-) Regards On Thu, Jan 20, 2011 at 7:15 PM, user virtual wrote: > Hi Guys > > I summarize all patches in one mail here because I am not certain each is > the best way to go, more like a discussion on how to finally close all holes > to support debian builds. > > Stefano provoked me to port all patches to unstable resulting in a big step > forward: > http://virtualusr.wordpress.com/2011/01/20/xen-4-x-testing-unstable-ubuntu/ > > Right! Almost all of the patches are already resolved in unstable > repository. For building xen-unstable debian packeges in my last attempt, > only one patch was absolutely necessary (debian-stubdom.Makefile.diff). > Strangely, manual run of "rules" makefile “fakeroot debian/rules binary” > builds everything even without this last patch applied, while using the > official packaging command "dpkg-buildpackage -rfakeroot" which calls the > above "rules" makefile fails without the patch. > It might be even theoretically possible to avoid this last patch, if one > could find which environment options applied by running dpkg-buildpackage > break the “make stubdom” command inside the "rules" makefile; find and unset > them. Unfortunately I have no time for this unsure guessing game now. > > I found one more bug (xen-watchdog-bug.diff) preventing the stopping of > xen-watchdog and by that provoking errors during the install and debian > packet management. > > While analyzing python patch (not on debian), I found out that it can be > dropped if I use the newer recommended debhelper command dh_pysupport to > pack python related parts. The issue however remains open (patch required) > if somebody only wishes to build sources without using packaging to debian > and dh_pysupport. > > Ian: - This special casing of Debian is pretty horrid. How about we > provide a general way to pass arguments to setup.py install and setup.py > build, with the variables set by default to the empty string? > > Sounds great Ian, but after looking into tools/python/setup.py, I feel this > exceeds my current knowledge of your python tools. > > Summary of attached patches against xen-unstable.hg: > 1. debian-stubdom.Makefile.diff - necessary for official packaging command > "dpkg-buildpackage -rfakeroot", not needed for simple build without .deb > packages > 2. xen-watchdog-bug.diff - obvious bug on all platforms > 3. debian-python-2.6.diff - only needed for simple build on debian with no > .deb packaging > *.err - describes the error resolved by the patch > > I would personally like 1. to be included in xen-unstable.hg, and of course > bug 2. resolved. > Personally, I don't find 3. critical any more, as I always avoid installing > software which is not .deb packaged, but there might be other opinions out > there. > > Otherwise, great work guys. I find xen-unstable.hg "almost fully > debian-compatible". > > Best regards > > > PS: Please note that there is an outstanding issue which prevents compiling > from xen-unstable.hg with "export debug=n". > I will summarize my latest experience regarding missing debian patches in > xen-4.0-testing.hg in my next mail. > > > On Tue, Jan 11, 2011 at 11:07 PM, Ian Jackson wrote: > >> user virtual writes ("Re: [Xen-devel] Xen build - Debian vs Redhat layout >> patch options"): >> > I was encouraged by Bruce to forward my patches upstream to make Xen >> sources >> > buildable from debian makefile debian/rules and for debian based target >> > systems. >> >> Thanks. >> >> > I tried to make my 3 of 4 (5) patches to xen40-testing.hg as non >> intrusive >> > as possible >> > >> > - backport-StdGNU.mk.diff (from xen-devel, already here) >> >> Thanks for this. Posting them all one after another in a single >> message is not the most convenient format for a submission; better >> would be one per email. >> >> And there is no need to send us patches we already have; it just >> causes confusion ... >> >> > - debian-sysconfig-default.diff (conditional Makefile expansion) >> >> We have machinery in xen-unstable.hg now which should cope with this, >> so I don't think we need that change. >> >> Stefano has dealt with most of these I think, but we still had >> outstanding this: >> >> >> > Description: Simple Python install layout for Debian based systems >> > Ported from origin below (Xen-3.4), which is a follower of >> > previous version from ubuntu Xen-3.3 sources >> > xen-3.3-3.3.0/debian/patches/use-python-2.6.diff >> > Drivers: User Virtual >> > URL: http://virtualusr.wordpress.com >> > Origin: >> https://bugs.launchpad.net/ubuntu/+source/xen-3.3/+bug/378240/+attachment/662345/+files/xen-3.4_3.4.1-1ubuntu1.diff.gz >> > Index: xen-4.0-4.0.1/tools/pygrub/Makefile >> > =================================================================== >> > --- xen-4.0-4.0.1.orig/tools/pygrub/Makefile 2010-08-25 >> 10:22:42.000000000 +0200 >> > +++ xen-4.0-4.0.1/tools/pygrub/Makefile 2011-01-04 >> 19:08:18.556582404 +0100 >> > @@ -11,7 +11,8 @@ >> > .PHONY: install >> > install: all >> > CC="$(CC)" CFLAGS="$(CFLAGS)" $(PYTHON) setup.py install \ >> > - $(PYTHON_PREFIX_ARG) --root="$(DESTDIR)" --force >> > + $(PYTHON_PREFIX_ARG) --root="$(DESTDIR)" --force \ >> > + $(if $(filter y,$(DEBIANBASE)),--install-layout=deb) >> > $(INSTALL_PYTHON_PROG) src/pygrub $(DESTDIR)/$(BINDIR)/pygrub >> > $(INSTALL_DIR) $(DESTDIR)/var/run/xend/boot >> >> This special casing of Debian is pretty horrid. How about we provide >> a general way to pass arguments to setup.py install and setup.py >> build, with the variables set by default to the empty string ? >> >> Ian. >> > >