| Commit message (Collapse) | Author | Age |
|\
| |
| |
| |
| | |
Conflicts:
dh
|
| |
| |
| |
| |
| |
| |
| |
| | |
* dh_auto_build, dh_auto_configure, dh: Set environment variables
listed by dpkg-buildflags --export. Any environment variables that
are already set to other values will not be changed.
Closes: #544844
* Also, support DEB_BUILD_OPTIONS=noopt, by changing -O2 to -O0.
|
|\|
| |
| |
| |
| |
| | |
Conflicts:
debhelper.pod
debian/changelog
|
| |
| |
| |
| |
| |
| |
| | |
--libexecdir when using autoconf. Closes: #541458
Fixed rleigh's patch to be more correct in the edge case where there is
a non-multiarch dpkg (ie, backports).
|
| | |
|
| |
| |
| |
| |
| | |
Appending @_ to a string appends the array length rather than
the array contents, so join with separating whitespace.
|
| | |
|
| |
| |
| |
| |
| |
| |
| |
| | |
The makefile parse causes dh to be run recursively.
Before, dh would just immediatly fail with "unknown sequence", but
now it has to run the makefile parse to calculate the sequences, so an
earlier bailout is needed.
|
| |
| |
| |
| |
| |
| | |
An empty explicit target in debian/rules should still be run,
to run its dependencies, and allow defining empty targets in order to
skip running what's nornally done by a sequence.
|
| | |
|
| | |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
targets
This assumes that all implicit rules file targets are owned by dh, so
it can just assume an implicit target can be optimized away to the commands
in its sequence.
I suppose this would break:
build:
dh build
install: build
dh install
binary-%: install
my-binary-builder-$@
my-binary-builder-arch:
echo "whee! I did something pointlessly complicated with make!"
dh binary-arch
my-binary-builder-indep:
dh binary-indep
But I can't imagine anyone does this, at least with a probability of 1 in
ten thousand, so hopefully it doesn't break any existing packages. :)
|
| |
| |
| |
| |
| |
| |
| | |
Reorder code so sequences can all be built before addons are loaded, so
addon interface can always affect all commands in any sequences. This fixes
a bug in the previous patch, where addons could not influence dh_testdir
and dh_testroot.
|
| | |
|
| |
| |
| |
| |
| |
| |
| |
| | |
Callers overriding build targets will need to configure by hand or by
calling dh_auto_configure, which should be the status quo. And moving
dh_auto_configure to build (and build-arch and build-indep) will not
make it run twice AFAICS (except for the edge case when it already did:
debian/rules build-arch build-indep)
|
| |
| |
| |
| |
| | |
These calls are no-ops, unless explicit targets exist, and in that case
later code rewrites the build and install sequences to include them.
|
| | |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The build and install rules run a minimal sequence if the build-arch or
build-indep, or install-arch or install-indep targets, respectively,
are present in debian/rules. The purpose is to not do work ahead of
time, such as building before the build-arch or build-indep targets are
built, which could potentially lead to misbuilds. If the targets are
not defined, the sequences may be run directly which is faster due to
being able to run the arch and indep commands together.
|
|/
|
|
|
|
|
|
|
| |
Rather than dh sequences containing dependent sequences within
themselves, invoke the sub-sequence via debian/rules to permit
overriding and customisation using the policy-defined debian/rules
targets.
Signed-off-by: Roger Leigh <rleigh@debian.org>
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
| |
While making postrm-ucf more like the other autoscripts I introduced two typo's
in the code. This patch fixes them.
|
| |
|
|
|
|
| |
files to be missed. Closes: #624377
|
| |
|
| |
|
|
|
|
| |
does not process files from ARGV
|
|
|
|
| |
does not actually install stuff
|
|
|
|
|
|
| |
Here is a patch against the master branch that adds such a command
called dh_installucf. It also registers the conffiles with ucfr and
removes stray ucf-{new,old,dist} files on purge.
|
| |
|
| |
|
| |
|
|
|
|
| |
being acted on. dh can sometimes produce such logs by accident when passed bundled options (like "-Nfoo" instead of "-N foo") that it does not understand; and it was not possible to fix that for any compat level before v8. But also, such logs can occur for other reasons, like interrupted builds during development, and it should be safe to clean them all. Closes: #623446
|
| |
|
| |
|
| |
|
|\ |
|
| | |
|
|/ |
|
|
|
|
| |
#604727
|
| |
|
|
|
|
|
|
| |
Note to backporters: If you remove that dependency, debhelper will fall
back to not doing multiarch stuff in v9 mode, which is probably what you
want.
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Open compat level 9, which incompatibly changes dh_auto_configure behavior
to set --libdir and --libexecdir to the multiarch directory path. This
requires dpkg-dev 1.16.0 (not yet released) for the multiarch directory
variable, so bump the dependency to this version.
Also set a new substvar, misc:Pre-Depends, to multiarch-support, a virtual
package provided by versions of eglibc that support the multiarch library
paths at runtime; this needs to be a pre-dependency to ensure unpacked but
not-yet-configured libraries can still be found during upgrades, so library
packages converting to multiarch (i.e., switching to compat 9) will need to
add this substitution by hand to debian/control.
|
|
|
|
| |
rule other than clean, perhaps to clean up an intermediate build before a second build is run, debian/rules clean still runs it. Closes: #615553
|
| |
|