David Kalnischkies [Sat, 3 Jul 2010 22:32:52 +0000 (00:32 +0200)]
prefer the Policy if it is built instead of the DepCache and
if DepCache is not available as fallback built the Policy
David Kalnischkies [Sat, 3 Jul 2010 21:55:12 +0000 (23:55 +0200)]
reorder classes a bit and make TryToInstallBuildDep use them
David Kalnischkies [Sat, 3 Jul 2010 18:09:17 +0000 (20:09 +0200)]
Refactor TryToInstall to look a bit saner by splitting the Remove
and the Virtual packages part out of the loop.
The function still exists unchanged as TryToInstallBuildDep through
for the BuildDep installation method
David Kalnischkies [Fri, 2 Jul 2010 05:06:53 +0000 (07:06 +0200)]
add a ConstructedBy member to the PackageSet which can be used by the
e.g. FromString to tell the caller if the string was an exact match
or found by regex or task. The two later ones can match packages for
which we want to ignore failures in the VersionSet
David Kalnischkies [Tue, 29 Jun 2010 17:21:35 +0000 (19:21 +0200)]
if the package has no installed & candidate but is virtual see if only
one package provides it - if it is only one use this package instead
David Kalnischkies [Tue, 29 Jun 2010 17:10:47 +0000 (19:10 +0200)]
rename AddSelectedVersion() to a better public FromPackage()
David Kalnischkies [Tue, 29 Jun 2010 15:14:45 +0000 (17:14 +0200)]
for install, do all installs first and then the removes and vice versa
David Kalnischkies [Mon, 28 Jun 2010 20:13:17 +0000 (22:13 +0200)]
- factor out code to get a single package FromName()
- check in Grouped* first without modifier interpretation
David Kalnischkies [Sat, 26 Jun 2010 19:17:34 +0000 (21:17 +0200)]
- only print errors if all tries to get a package by string failed
*
David Kalnischkies [Sat, 26 Jun 2010 18:49:47 +0000 (20:49 +0200)]
always mark the all package if a pseudo package is marked for install
David Kalnischkies [Sat, 26 Jun 2010 17:22:23 +0000 (19:22 +0200)]
give the APT::Cache::Generate option her effect back
David Kalnischkies [Sat, 26 Jun 2010 17:04:20 +0000 (19:04 +0200)]
do not override the user set quiet setting even if the target is not a tty
David Kalnischkies [Sat, 26 Jun 2010 16:42:42 +0000 (18:42 +0200)]
make the MMap Grow Error a fatal one as while in theory the code should
never segfault it still tend to do it so better show it directly
David Kalnischkies [Sat, 26 Jun 2010 16:40:15 +0000 (18:40 +0200)]
* apt-pkg/contrib/fileutl.cc:
- show notice about ignored file instead of being always silent
- add a Dir::Ignore-Files-Silently list option to control the notice
*
David Kalnischkies [Sat, 26 Jun 2010 11:29:24 +0000 (13:29 +0200)]
- use the new MatchAgainstConfig for the DefaultRootSetFunc
* apt-pkg/contrib/configuration.{cc,h}:
- add a wrapper to match strings against configurable regex patterns
David Kalnischkies [Sat, 26 Jun 2010 07:03:26 +0000 (09:03 +0200)]
* apt-pkg/aptconfiguration.cc:
- show a deprecation notice for APT::Acquire::Translation
David Kalnischkies [Fri, 25 Jun 2010 18:11:11 +0000 (20:11 +0200)]
print all messages if the application is in an interactive run
David Kalnischkies [Fri, 25 Jun 2010 17:16:12 +0000 (19:16 +0200)]
add a simple stack handling to be able to delay error handling
David Kalnischkies [Fri, 25 Jun 2010 06:01:48 +0000 (08:01 +0200)]
* apt-pkg/contrib/error.{cc,h}:
- complete rewrite but use the same API
- add NOTICE and DEBUG as new types of a message
David Kalnischkies [Tue, 22 Jun 2010 16:01:11 +0000 (18:01 +0200)]
Use an abstract helper for error handling and output
instead of doing this directly in the CacheSets.
With this method an application like apt-get can change
the behavior of the CacheSets to his liking.
It can for example easily keep track of how packages
were added to the set: by names or with regex's…
David Kalnischkies [Sat, 19 Jun 2010 12:16:40 +0000 (14:16 +0200)]
* cmdline/apt-get.cc:
- use the cachsets in the install commands
David Kalnischkies [Sat, 19 Jun 2010 09:49:38 +0000 (11:49 +0200)]
get packages by task^ with FromTask()
David Kalnischkies [Sat, 19 Jun 2010 07:13:33 +0000 (09:13 +0200)]
fix the gcc warning about the initialisation order of variables caused
by moving Policy to public again (and therefore after SrcList)
David Kalnischkies [Wed, 16 Jun 2010 15:48:37 +0000 (17:48 +0200)]
* apt-pkg/depcache.cc:
- SetCandidateVer for all pseudo packages
- SetReInstall for the "all" package of a pseudo package
David Kalnischkies [Wed, 16 Jun 2010 15:19:21 +0000 (17:19 +0200)]
add a simple method to VerIterator to check if two Versions are similar
David Kalnischkies [Tue, 15 Jun 2010 18:46:09 +0000 (20:46 +0200)]
* cmdline/cacheset.cc:
- doesn't include it in the library for now as it is too volatile
David Kalnischkies [Tue, 15 Jun 2010 18:29:50 +0000 (20:29 +0200)]
Add a GroupedFromCommandLine for the VersionSet similar to the
one for PackageSet and refactor the existing VersionSet methods
to simplify that.
David Kalnischkies [Fri, 11 Jun 2010 06:51:01 +0000 (08:51 +0200)]
* debian/control:
- remove intltool's dependency as it is an ubuntu artefact
David Kalnischkies [Thu, 10 Jun 2010 14:37:18 +0000 (16:37 +0200)]
Merge ... debian-experimental-ma to get the public policy back
Michael Vogt [Thu, 10 Jun 2010 14:15:57 +0000 (16:15 +0200)]
* apt-pkg/cachefile.h:
- make pkgPolicy public again, libapt-pkg-perl (and probably
others) get unhappy without that
David Kalnischkies [Thu, 10 Jun 2010 13:10:38 +0000 (15:10 +0200)]
* apt-pkg/orderlist.cc:
- untouched packages are never missing
* apt-pkg/packagemanager.cc:
- packages that are not touched doesn't need to be unpacked
Michael Vogt [Thu, 10 Jun 2010 12:35:23 +0000 (14:35 +0200)]
releasing version 0.7.26~exp6
Michael Vogt [Thu, 10 Jun 2010 12:01:18 +0000 (14:01 +0200)]
merged from david
David Kalnischkies [Thu, 10 Jun 2010 11:09:23 +0000 (13:09 +0200)]
* apt-pkg/cacheset.cc:
- get the candidate either from an already built depcache
or use the policy which is a bit faster than depcache generation
David Kalnischkies [Thu, 10 Jun 2010 08:03:14 +0000 (10:03 +0200)]
correct a minor spelling mistake in the changelog
David Kalnischkies [Thu, 10 Jun 2010 00:04:03 +0000 (02:04 +0200)]
readd the autoremove protection for the kfreebsd-image as it was lost
in Michaels merge with the ubuntu branch
David Kalnischkies [Wed, 9 Jun 2010 21:20:26 +0000 (23:20 +0200)]
enhance the split out of the gpgv commandline mangling by splitting out
the call completely
David Kalnischkies [Wed, 9 Jun 2010 19:15:52 +0000 (21:15 +0200)]
sent Last-Modified header also for Translation files
David Kalnischkies [Wed, 9 Jun 2010 16:23:23 +0000 (18:23 +0200)]
be sure that the RFC1123StrToTime method is run in a LANG=C environment
David Kalnischkies [Wed, 9 Jun 2010 15:19:40 +0000 (17:19 +0200)]
* apt-pkg/indexcopy.cc:
- move the gpg codecopy to a new method and use it also in methods/gpgv.cc
David Kalnischkies [Wed, 9 Jun 2010 15:13:40 +0000 (17:13 +0200)]
fix compiler warning in the new mirror code
David Kalnischkies [Wed, 9 Jun 2010 12:20:27 +0000 (14:20 +0200)]
* apt-pkg/policy.cc:
- get the candidate right for a not-installed pseudo package if
his non-pseudo friend is installed
David Kalnischkies [Wed, 9 Jun 2010 12:18:56 +0000 (14:18 +0200)]
remove the "typename" from the const_iterators as gcc-4.4 doesn't like
them and gcc-4.5 does the right thing (TM) with and without them
Michael Vogt [Wed, 9 Jun 2010 11:15:34 +0000 (13:15 +0200)]
merged from lp:~donkult/apt/sid
Michael Vogt [Wed, 9 Jun 2010 11:03:41 +0000 (13:03 +0200)]
debian/rules: remove two more leftovers fromthe ubuntu merge
Michael Vogt [Wed, 9 Jun 2010 10:08:54 +0000 (12:08 +0200)]
merge lp:~mvo/apt/ubuntu-mirror-method-improvements
Michael Vogt [Wed, 9 Jun 2010 09:51:21 +0000 (11:51 +0200)]
* merge the remaining Ubuntu change:
- on gpg verification failure warn and restore the last known
good state
- on failure display the IP of the server (useful for servers
that use round robin DNS)
- support Original-Maintainer in RewritePackageOrder
- enable cdrom autodetection via libudev by default
- show messsage about Vcs in use when apt-get source is run for
packages maintained in a Vcs
- better support transitional packages with mark auto-installed.
when the transitional package is in "oldlibs" the new package
is not marked auto installed (same is true for section
metapackages)
- provide new "deb mirror://archive.foo/mirrors.list sid main"
method expects a list of mirrors (generated on the server e.g.
via geoip) and will use that, including cycle on failure
- write apport crash file on package failure (disabled by default
on debian until apport is available)
- support mirror failure reporting (disabled by default on debian)
David Kalnischkies [Wed, 9 Jun 2010 08:52:45 +0000 (10:52 +0200)]
- backport forgotten Valid-Until patch from the obsolete experimental
branch to prevent replay attacks better, thanks to Thomas Viehmann
for the initial patch! (Closes: #499897)
* doc/apt.conf.5.xml:
- document the new Valid-Until related options
* apt-pkg/contrib/strutl.cc:
- split StrToTime() into HTTP1.1 and FTP date parser methods and
use strptime() instead of some self-made scanf mangling
- use the portable timegm shown in his manpage instead of a strange
looking code copycat from wget
* ftparchive/writer.cc:
- add ValidTime option to generate a Valid-Until header in Release file
David Kalnischkies [Wed, 9 Jun 2010 08:50:32 +0000 (10:50 +0200)]
* apt-pkg/pkgcache.h:
- switch {,Install-}Size to unsigned long long
- deal with long long, not with int to remove 2GB Limit (LP: #250909)
- deprecate AddSize with Multiplier as it is unused and switch to
boolean instead to handle the sizes more gracefully.
- switch i{Download,Usr}Size from double to (un)signed long long
* cmdline/apt-get.cc:
- use unsigned long long instead of double to store values it gets
David Kalnischkies [Wed, 9 Jun 2010 08:46:35 +0000 (10:46 +0200)]
rename the options, document them and reorder the changelog a bit
David Kalnischkies [Tue, 8 Jun 2010 22:53:44 +0000 (00:53 +0200)]
move the users away from the deprecated StrToTime() method
David Kalnischkies [Tue, 8 Jun 2010 22:27:22 +0000 (00:27 +0200)]
use the portable timegm shown in his manpage instead of a strange
looking code copycat from wget
David Kalnischkies [Tue, 8 Jun 2010 22:12:14 +0000 (00:12 +0200)]
* ftparchive/writer.cc:
- add ValidTime option to generate a Valid-Until header in Release file
David Kalnischkies [Tue, 8 Jun 2010 21:38:34 +0000 (23:38 +0200)]
use the Valid-Until header from the Release file but if the user
provides a setting in the configuration prefer the date which is
earlier.
David Kalnischkies [Tue, 8 Jun 2010 20:55:42 +0000 (22:55 +0200)]
enhance the Valid-Until code a bit by using the correct RFC1123StrToTime
method and allow for better translations of the error messages
David Kalnischkies [Tue, 8 Jun 2010 20:46:42 +0000 (22:46 +0200)]
* apt-pkg/contrib/strutl.cc:
- split StrToTime() into HTTP1.1 and FTP date parser methods and
use strptime() instead of some selfmade scanf mangling
David Kalnischkies [Tue, 8 Jun 2010 17:27:49 +0000 (19:27 +0200)]
* apt-pkg/indexrecords.cc:
- backport forgotten Valid-Until patch from the obsolete experimental
branch to prevent replay attacks better, thanks to Thomas Viehmann
for the initial patch! (Closes: #499897)
David Kalnischkies [Sun, 6 Jun 2010 19:49:28 +0000 (21:49 +0200)]
use the GroupedFromCommandLine() method in the dotty and xvcg command
to get all the funky features in less lines
David Kalnischkies [Sun, 6 Jun 2010 19:41:27 +0000 (21:41 +0200)]
add a GroupedFromCommandLine method to the PackageSet to split the
packages on the commandline into groups based on modifiers
David Kalnischkies [Sat, 5 Jun 2010 17:14:32 +0000 (19:14 +0200)]
do not fail if an unrelated error is pending in DisplayRecord()
David Kalnischkies [Sat, 5 Jun 2010 16:00:00 +0000 (18:00 +0200)]
support special release-modifier 'installed' and 'candidate'
David Kalnischkies [Sat, 5 Jun 2010 14:59:45 +0000 (16:59 +0200)]
accept package versions in the unmet command
David Kalnischkies [Sat, 5 Jun 2010 14:45:19 +0000 (16:45 +0200)]
enhance the changelog a bit
David Kalnischkies [Sat, 5 Jun 2010 13:47:16 +0000 (15:47 +0200)]
handle :arch modifier in PackageSet::FromString correctly
David Kalnischkies [Sat, 5 Jun 2010 11:42:04 +0000 (13:42 +0200)]
apt-cache show --no-all-versions should issues the Candidate
David Kalnischkies [Sat, 5 Jun 2010 11:38:17 +0000 (13:38 +0200)]
do not insert end() Iterators into the CacheSet even if requested
David Kalnischkies [Sat, 5 Jun 2010 11:09:18 +0000 (13:09 +0200)]
Don't increase the commandline parameter in the library but in the
application to be really generic.
David Kalnischkies [Sat, 5 Jun 2010 10:54:30 +0000 (12:54 +0200)]
add --target-release option (Closes: #115520)
David Kalnischkies [Sat, 5 Jun 2010 08:17:10 +0000 (10:17 +0200)]
Add a option to apt-cache policy to additionally init the DepCache before
starting to get the package informations.
This is useful e.g. for debugging the MultiArchKiller.
David Kalnischkies [Sat, 5 Jun 2010 08:04:48 +0000 (10:04 +0200)]
use pkgCacheFile and the new CacheSets all over the place
David Kalnischkies [Sat, 5 Jun 2010 07:20:29 +0000 (09:20 +0200)]
Add a method to get a VersionSet from the Commandline and refactor
the existing methods a bit to reuse them easier intern
David Kalnischkies [Fri, 4 Jun 2010 20:58:57 +0000 (22:58 +0200)]
store also the SourceList we use internally for export
David Kalnischkies [Fri, 4 Jun 2010 12:51:21 +0000 (14:51 +0200)]
Switch away from the now deprecated methods for Cache building
David Kalnischkies [Fri, 4 Jun 2010 12:43:03 +0000 (14:43 +0200)]
* apt-pkg/cachefile.{cc,h}:
- split Open() into submethods to be able to build only parts
- make the OpProgress optional in the Cache buildprocess
David Kalnischkies [Thu, 3 Jun 2010 08:48:19 +0000 (10:48 +0200)]
* cmdline/apt-get.cc:
- use unsigned long long instead of double to store values it gets
David Kalnischkies [Thu, 3 Jun 2010 08:45:58 +0000 (10:45 +0200)]
* apt-pkg/depcache.cc:
- switch i{Download,Usr}Size from double to (un)signed long long
The biggest reason is that this saves a lot of float point operations
we do in AddSizes() on integers. The only reason i see that this was
a double is that it was 64bit long and can therefore store bigger
values than int/long, but with the availablity of (un)signed long long
we are now also at 64bit and can store sizes more than 8 Exabytes big -
by the time this will be a limit the C/C++ Standard will have bigger types,
hopefully.
David Kalnischkies [Thu, 3 Jun 2010 07:50:06 +0000 (09:50 +0200)]
deprecate AddSize with Multiplier as it is unused and switch to
boolean instead to handle the sizes more gracefully.
David Kalnischkies [Thu, 3 Jun 2010 07:22:00 +0000 (09:22 +0200)]
* apt-pkg/pkgcache.h:
- switch {,Install-}Size to unsigned long long
* apt-pkg/depcache.cc:
- deal with long long, not with int to remove 2GB Limit (LP: #250909)
Michael Vogt [Wed, 2 Jun 2010 21:57:00 +0000 (23:57 +0200)]
methods/mirror.cc: debug improvements
Michael Vogt [Wed, 2 Jun 2010 21:30:01 +0000 (23:30 +0200)]
implement Fail-Ignore bool in FetchItem that tells the method that a failure of this item is ok and does not need to be tried on all mirrors
Michael Vogt [Wed, 2 Jun 2010 21:15:55 +0000 (23:15 +0200)]
methods/mirror.cc: simplify uri.startswith()
Michael Vogt [Wed, 2 Jun 2010 21:13:42 +0000 (23:13 +0200)]
methods/mirrors.cc: make cycle through the mirrors work properly
Michael Vogt [Wed, 2 Jun 2010 20:47:49 +0000 (22:47 +0200)]
methods/mirror.cc: remove Acquire::Mirror::RefreshInterval (not really useful)
Michael Vogt [Wed, 2 Jun 2010 20:44:49 +0000 (22:44 +0200)]
improve error message if mirror method fails
Michael Vogt [Wed, 2 Jun 2010 19:43:55 +0000 (21:43 +0200)]
cmdline/apt-report-mirror-failure: port to latest python-apt
Michael Vogt [Wed, 2 Jun 2010 14:51:25 +0000 (16:51 +0200)]
methods/mirror.cc: on fail try the next mirror
Michael Vogt [Wed, 2 Jun 2010 14:40:40 +0000 (16:40 +0200)]
* apt-pkg/deb/dpkgpm.cc:
- add missing include
* methods/mirror.{cc,h}:
- add SelectNextMirror() and InitMirrors() functions
- read all mirrors into the AllMirrors vector
David Kalnischkies [Mon, 31 May 2010 20:36:41 +0000 (22:36 +0200)]
* apt-pkg/cacheset.{cc,h}:
- add simple wrapper around std::set for cache structures
David Kalnischkies [Mon, 31 May 2010 19:00:09 +0000 (21:00 +0200)]
rename packageset into cacheset while it is not too late
David Kalnischkies [Mon, 31 May 2010 18:02:41 +0000 (20:02 +0200)]
* cmdline/apt-cache.cc:
- make the search multiarch compatible by using GrpIterator instead
David Kalnischkies [Mon, 31 May 2010 18:00:16 +0000 (20:00 +0200)]
* apt-pkg/pkgcache.cc:
- get the best matching arch package from a group with FindPreferredPkg
David Kalnischkies [Mon, 31 May 2010 15:22:04 +0000 (17:22 +0200)]
* apt-pkg/contrib/cmdline.cc:
- fix segfault in SaveInConfig caused by writing over char[] sizes
David Kalnischkies [Mon, 31 May 2010 12:16:44 +0000 (14:16 +0200)]
improve the handling of regex together with the architecture modifier
David Kalnischkies [Mon, 31 May 2010 10:58:20 +0000 (12:58 +0200)]
i managed to commit broken code… which (at least in my mind) worked
yesterday. Strange… anyway only small fixes.
David Kalnischkies [Sun, 30 May 2010 22:28:14 +0000 (00:28 +0200)]
move cmdline parsing from apt-cache to new FromCommandLine method
David Kalnischkies [Sun, 30 May 2010 21:12:41 +0000 (23:12 +0200)]
move regex magic from apt-get to new FromRegEx method
David Kalnischkies [Sun, 30 May 2010 05:57:32 +0000 (07:57 +0200)]
* apt-pkg/packageset.h:
- add a simple wrapper around std::set for packages with it
David Kalnischkies [Sat, 29 May 2010 17:10:19 +0000 (19:10 +0200)]
* ftparchive/writer.h:
- add a virtual destructor to FTWScanner class (for cppcheck)
David Kalnischkies [Sat, 29 May 2010 17:02:32 +0000 (19:02 +0200)]
* apt-pkg/depcache.cc:
- correct "Dangerous iterator usage." pointed out by cppcheck
David Kalnischkies [Sat, 29 May 2010 13:44:47 +0000 (15:44 +0200)]
* apt-pkg/cacheiterator.h:
- let pkgCache::Iterator inherent std::iterator