| Age | Commit message (Collapse) | Author |
|---|
|
As outlined in #748355 apt segfaulted if it encountered a loop between a
package pre-depending on a package conflicting with the previous as it
ended up in an endless loop trying to unpack 'the other package'.
In this specific case as an essential package is involved a lot of force
needs to be applied, but can also be caused by 'normal' tight loops and
highlights a problem in how we handle breaks which we want to avoid.
The fix comes in multiple entangled changes:
1. All Smart* calls are guarded with loop detection. Some already had it,
some had parts of it, some did it incorrect, and some didn't even try.
2. temporary removes to avoid a loop (which is done if a loop is
detected) prevent the unpack of this looping package (we tried to unpack
it to avoid the conflict/breaks, but due to a loop we couldn't, so we
remove/deconfigure it instead which means we can't unpack it now)
3. handle conflicts and breaks very similar instead of duplicating most
of the code. The only remaining difference is, as it should:
deconfigure is enough for breaks, for conflicts we need the big hammer
|
|
Git-Dch: Ignore
|
|
In bugreport #747261 I confirmed with this testcase that apt actually
supports the requested architecture-specific conflicts already since
2012 with commit cef094c2ec8214b2783a2ac3aa70cf835381eae1.
The old test only does simulations which are handy to check apt,
this one builds 'real' packages to see if dpkg agrees with us.
Git-Dch: Ignore
|
|
Version/Architecture information in a Translation-$lang file is
not allowed, so don't try to parse it. This is a fix for a bugreport
where a Translation-en file contained the content of the regular
Packages file (probably due to local FS corruption). This lead to
strange error messages on file download.
Thanks to Thomas Reusch for the report.
|
|
|
|
debian/sid
|
|
Closes: 746434
|
|
The bugreport highlights the problem with an empty package name. We fix
this by 'ignoring' these so that it behaves just like "apt-get install".
The deeper problem is that modifier strings can be longer than a package
name in which case the comparison doesn't make sense, so don't compare
then. Was not noticed so far as all modifiers are of length 1, so the
only package name shorter than this is in fact the empty package name.
Closes: 744940
|
|
Conflicts:
test/integration/test-apt-cli-list
|
|
|
|
A package which can't be downloaded anymore is very likely dropped from
a release and can therefore no longer be 'standard' (or similar). We
therefore do not grant points for them anymore and demote them to
prio:extra instead which helps other packages breaking them away even if
they have a lower priority.
The testcase was initially created by Michael Vogt and just amended.
|
|
|
|
If the user is using "apt list pattern" and there is only a single
hit, notice about "--all-versions" as this is what the user may
be interessted in
|
|
This patch should fix spurious test failures in jenkins or travis
that are caused by a race condition in the {stunnel,aptwebserver}.pid
file creation
|
|
Packages in the "deinstall ok config-file" have no candidate or
instaleld version. So they must be special cased in the apt list
generation.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Use mkstemp() in apt-extractemplates and add a integrationtest
for apt-extracttemplates too. Thanks to Steve Kemp for the report.
|
|
In commit 446551c8 I changed MarkInstall to discard the candidate if the
candidate can't satisfy the dependency. This breaks interactive solvers
like aptitude which can change the candidate on-the-fly later.
In commit df77d8a5 I introduced this 'early' loop-breaking to begin with
which can't be that helpful for interactive solvers as well, but makes
perfect sense for non-interactives to stop them from exploring trees
which can't be satisfied, but it isn't perfect as ideally we would check
this before auto-installing the first dependency.
This commit therefore moves the loop into its own IsInstallOk hook so
that frontends can override this check if they want to and in exchange
removes the loop-breaking from MarkInstall itself and does it before any
dependency is installed.
Closes: 740750
|
|
We have to properly close our pseudo terminals even in error cases
before we call post-invoke scripts. This is done now by breaking from
the dpkg calling loop instead of copying the handling, which did it in
the wrong order before.
This also ensures that our state file is written in error cases to
record autobit and co as this was forgotten before.
Closes: 738969
|
|
Git-Dch: Ignore
|
|
Metapackages like "linux-image-amd64" are otherwise matched by our
extraction as well, which later on can't be successfully compared via
dpkg --compare-versions as the 'amd64' bit isn't a version number.
(Luckily none of our architectures starts with a digit.)
This was broken by me in 0.9.16 as I moved a shell-glob matcher to a
regex-based one which has slightly different semantics regarding '*'.
Closes: 741962
|
|
|
|
The framework can be configured to use different compression algorithms
to test different ones, but a testcase testing for gz support should
always be run with gz, regardless of what compressions are configured
otherwise.
Git-Dch: Ignore
|
|
Beside fixing this minor code duplication it also resolves the problem
of messing up vim syntax-highlighting.
Git-Dch: Ignore
|
|
Mostly ensures that we use the build methods and not the system
provided methods in the tests (if we don't want it that way).
Git-Dch: Ignore
|
|
As we deal with regex matchers here the dots are treated as wildcards if
we don't take care of escaping them. Not very likely that this could be
a real-world problem, but just to be sure.
|
|
kfreebsd as well as hurd kernel packages call the postinst script as
well so we just need to enable the correct parsing for installed
packages and disable the "protect every version" hammer for them.
|
|
With APT::VersionedKernelPackages users have the option of adding
packages like pre-build out-of-tree modules to the list of automatically
protected from being autoremoved.
|
|
Git-Dch: Ignore
|
|
fixes some messages and their translation so that all of them have three
dots for messages with an elipse. Many translations already had this.
|
|
Git-Dch: Ignore
|
|
Git-Dch: Ignore
|
|
The unpack of a M-A:same package will force the unpack of all its
siblings directly to prevent that they could be separated by later
immediate actions. In commit 634985f8 a call to SmartConfigure was
introduced to configure these packages at the time the installation
order encounters them. Usually, the unpack order is already okay, so
that this 'earlier' unpack was not needed and if it wouldn't have been
done, the package would now only be unpacked, but by configuring the package
now we impose new requirements which must be satisfied. The code is
clever enough to handle this most of the time (it worked for 2 years!),
but it isn't needed and in very coupled cases this can fail.
Removing this call again removes this extra burden and so simplifies the
ordering as can be seen in the modified tests. Famous last words, but I
don't see a reason for this extra burden to exist hence the remove.
Closes: 740843
|
|
Git-Dch: Ignore
|
|
Inspired by the rest of the patch in 661537, but abstract the
parsing of various ways of setting the build profiles more so it can
potentially be reused and all apt parts have the same behaviour.
Especially config options, cmdline options and environment will not be
combined as proposed as this isn't APTs usual behaviour and dpkg doesn't
do it either, so one overrides the other as it normally does.
|
|
Prevents that "old" dependencies have an influence in the scoring.
With positive dependencies this is usually not a problem, but negative
dependencies can linger around for a long time.
|
|
Git-Dch: Ignore
|
|
versioned -dev packages like db and boost have the problem of no
dependencies which would give them a competitive advantage against an
older incarnation of the -dev package, so they tend to be kept back
until the old version is removed from the archive, which, if the user
has older releases in its sources can take a long time (or never happens).
The newer version has a conflicts/breaks against the older one, but the
older one hasn't against the newer, so by giving via the conflicts the
older one a reduced score the newer one can win if there is no other
reason to keep it. If both have a conflict against each other the
scoring will cancel itself out, so no harm done.
This gives "action" a slightly bigger edge in breaks/conflicts cases
than before, but holding back isn't a really good solution anyway.
|
|
feature/apt-download-file
Conflicts:
cmdline/apt-helper.cc
test/integration/framework
test/integration/test-apt-https-no-redirect
|
|
|
|
|
|
|
|
http-method from METHODSDIR
|
|
|
|
|
